A big woof from your favourite interspecies team!

Wins -

Our paper went live in the Journal of Clinical Oncology (JCO) at the end of April. JCO is the leading global oncology journal that clinicians around the world read to update their practice. JCO commissioned an exclusive editorial to highlight our paper (reserved for 5% of articles) and invited us to their podcast (1/month) to discuss our results of 90%+ accuracy in detecting 20 types of cancers from breath, including 90%+ sensitivity at Stage-1.

Following the publication, as well as announcement of being part of Accel India and Prosus’ AtomsX, Dognosis received a ton of organic press from BBC News, Bloomberg, WION, NDTV Profit, CNBC-TV18, Times of India, India Today, Business Standard, Hindistan Times, TechCrunch, YourStory, Inc42, Startuppedia, Indian Express, STAT, Happier Health and many more. Some highlights:

• Rahul, co-founder of the premier Indian long-form media-house Tigerfeathers wrote a wonderful piece on our story and vision, and Caleb did a great in-depth video at Runtime BRT.

• Sadhguru, the global icon with millions of followers, spoke about us in one of his weekly talks.

• Packy from Not Boring covered us in a Weekly Dose of Optimism.

• We were invited to give a talk and panel, and be part of a documentary at AI4.co, America’s largest AI conference.

• Various accounts across Instagram, Linkedin, X, and Whatsapp made organic content (idkshowdaily, tech creator, 100xengineers) covering our work, aggregating over 1M views, and leading to more than 5000 new followers and 200k impressions on our socials.

This press and public interest translated to over 500 inbound leads for people interested in taking our test. We’ve done dozens of lengthy calls with these leads, generating rich insights on who are our early adopters are, who include:

• People with family/personal history of cancer

• 20-30 year-olds who are interested in purchasing our test for their parents

• Dog-lovers who say they want to do our test specifically because they love our mission and how we’re unleashing the superpowers of dogs

The insights from these calls prompted us to rethink our GTM, and we’re doubling down on D2C as the main channel for our 2027 Q1 launch of BreathEasy in Bangalore.

The insight from these calls changed our plans. We’re now going straight to people instead of only through hospitals, starting with our BreathEasy launch in Bangalore in January 2027.

We’re onboarding top global cancer detection researchers and clinicians to our scientific and trial advisory board and we’re formalizing collaborations with leading global clinical institutes- more to come out soon.

Our team has only gotten stronger, with Sagar (India GTM Lead), Susmita (Regulatory Lead), Manal (Comms Lead), Madhu (Senior Data Scientist), Christy (People and Office Ops Manager), Mamatha (Lab Technician), Hari (Office Admin), Samarth (Associate Founders Office) and David (SWE with 4yrs of experience) joining the Pack, along with our friendly canine AI agent Lucky.

We’re hiring for multiple roles - including a Medical Affairs Lead (Doctor x Product), a Brand Marketer, Founding Designer, and a Head of Finance/Ops. We’d also love to hear from current or former founders and marketers with real experience building a D2C brand, and from people who understand how clinical imaging chains work, their incentives and structures, so we can build strong partnerships ahead of our commercial launch.

R&D -

SniffSpace V2 is officially in production. After months of design, prototyping, and testing, all our dogs have now moved fully onto V2 from V1, and production has a way of surfacing what got missed in development. We caught and fixed bugs across mechanical, electrical, and software-hardware integration, closed some safety gaps, and improved reliability, all without taking the platform offline. V2 brings automated sample presentation, real-time sample ID, sniff detection, localized VOC clearing between samples, and a vision-model-based reward system that automates another piece of the training loop. Overall, V2 has the capacity to improve our throughput to 288 sniffs-per-dog-hour. Next up: hardening the platform for scale, and preparing to deploy SniffspaceV2 to MDD (Medical Detection Dogs), our UK collaborator.

Our science team spent time stress-testing everything that could throw off a result before it ever reaches our dogs: different hospitals, different storage conditions, different collection settings, even a patient’s own stress levels. Alongside that, we tightened how the lab runs day to day, and started weaving in new signals, like a dog’s breathing and brain activity, to make every report stronger.

Plans -

Q3 is focused on bringing our systems to production-grade stability, executing on-going trials to generate case-studies of detecting undiagnosed cancers, and continuing to build our GTM, Marketing, and Medical Affairs functions. We’ll also aim to finalize our follow-up protocol, secure 2 partnerships with imaging chains with a great user experience, and have over a 1000 people on our product waitlist.

Parting Barks -

This has been the busiest things have ever been. The team has needed to execute on multiple fronts with a large amount of context-switching in a given day, time-frame shifting in every week, and dealing with all the growing pains of being a large organization with mounting stakers and complexity. It’s been helpful to have clear North Star goals for the remainder of this year:

• 1,000 samples a week by October,

• 5 undiagnosed early-stage cancers detected by December,

And of course, no matter how challenging it gets, both of us know we would never choose another timeline where we were not giving it our all to build Dognosis.

Dogspeed,

Akash and Itamar

The Dognosis Pack

Fifteen years passed before the question was formally tested by a team in the UK that published the first peer-reviewed study to show that trained dogs could detect cancer from a biological sample with statistically significant accuracy. Over the next two decades, studies followed across cancer types and sample modalities – breath, urine, blood, and tissue, many reporting sensitivities and specificities above 90%. The evidence kept mounting, beyond cancer to Parkinson’s, diabetes, COVID, disease after disease, to the growing realization that disease, almost any disease, leaves a chemical trace that a dog can find.

So why did this superpower never cohere into something medicine could use? The answer starts with what the dogs are actually detecting: volatile organic compounds (VOCs), the metabolic byproducts whose profile shifts when cancer disrupts normal cellular metabolism. Human breath alone contains over 1,400 distinct VOCs, and the dog’s nose can identify patterns within that complexity at parts per trillion, a sensitivity our best analytical instruments, working at parts per million, have never come close to matching.

And yet a nose, however extraordinary, cannot scale on its own. To do that, you need a way to read what the dog is detecting, in a form standardised and reproducible enough to screen populations rather than dozens, and robust enough for medicine to trust. In other words, we had to listen better. Which is why we built everything around the dogs, with their nose at the centre, including the technology to decode what they find. The results were published in the Journal of Clinical Oncology in April 2026.

The study and the numbers
We enrolled 3,275 participants across six hospitals in Karnataka, three in Hubballi and three in Bengaluru. Some had biopsy-confirmed cancer spanning seven cancer types, recruited before any treatment began. Others were healthy, or living with non-cancerous conditions, recruited from the same clinical settings so the dogs were responding to the scent of cancer, not the scent of a hospital. Each person did something simple: they breathed normally for ten minutes through a surgical mask. The mask was then sealed, logged, and sent to our lab – where seven of our superheroes – Chloe, Cheetah, Iggy, Billy, Whiskey, Jessie, and Bane took over.

Inside the lab, each session runs on SniffSpace, a custom eight-port platform where samples are presented in randomised arrays, barrier gates control access timing, QR scanners log every sample, and infrared sensors record the exact timestamp of each sniff.

Each dog gives a simple read: alert or pass, for each sample, and at least three dogs independently assessed every participant. Those individual reads were combined through a Bayesian fusion model that weighted each dog’s indication against their own historical sensitivity and specificity, and factored in each participant’s risk profile, arriving at a single calibrated cancer risk score. When we tested the system on 1,502 participants the dogs had never encountered before, this is what we found.

Sensitivity: 90.8%. Specificity: 91.3%. AUC: 0.962.

Consistent. Across the board.
Performance held across all seven cancer types: head and neck, breast, upper and lower GI, gynecologic, lung, and genitourinary, and critically, across disease stages. Sensitivity for Stage 1 and 2 was 90.6%, nearly identical to late-stage performance, which matters because the stage at which cancer is found determines almost everything about what happens next.

We tested the system across the variables that matter when you are screening millions of people, not hundreds. We simulated a dog having an off day, removing one dog from the ensemble entirely, and the system held. The result did not depend on any single dog. Performance did not change meaningfully by participants’ sex, tobacco history, or common comorbidities like diabetes and hypertension. It was stable regardless of whether they had eaten recently or come in for testing in the morning or afternoon.

It is also worth understanding what is driving the accuracy. A model built on participant-level clinical risk factors alone gave an AUC of 0.832, the dogs alone gave 0.932. Together, through the Bayesian fusion model, we reached 0.962, a result neither the dogs nor the clinical data could reach alone.

Teaching machines to listen
The Bayesian model decodes what the dogs find. What comes next is learning to read the dogs themselves more deeply and clearly than any human trainer could.

In this study, computer vision captured sniffing behaviour and dynamics at each port, tracking how long each dog lingered, the spatial path of their approach, and the rhythm of each sniff, feeding those signals into the Bayesian model alongside the dogs’ binary indications. Facial cameras picked up subtler signals still, ear position, eyelid movement, the barely perceptible shift in a dog’s expression when something catches their attention. In a pilot subset of 425 participants, this vision-augmented model achieved an AUC of 0.938, the first time automated behavioural sensing had been integrated into the fusion framework. The result suggests that automated sensing could eventually replace the human in the loop entirely, and that is exactly what we are building.

SniffSpace, our detection platform itself is being upgraded, moving from manual sample handling to a fully automated system where a robotic arm opens, presents, seals, and cleans each port between samples. Throughput jumps from eight samples per session to over 200, with three swappable tray sets running in rotation, RFID tags tracking every sample slot in real time, the dogs moving through their session, the platform rotating samples, the software logging every sniff event, with minimal human intervention.

DogSense, our patented brain-computer interface, is the next layer. Each helmet is individually sculpted to the dog’s head, fitting snugly enough to capture neurophysiological signals without interfering with the dog’s natural movement. What it reads in real time, brain activity, respiration, heart rate, interpreted by DogOS, our suite of machine learning models, captures signals no camera can. The goal is to make canine olfaction fully legible, not just whether a dog alerts, but what the alert means, how certain the dog is, and what is happening in their brain when they find it.

Verdict from the field
The Journal of Clinical Oncology is where oncologists go to update how they think and treat. In the issue our study was published, Dr Jonathan Friedberg, the journal’s Editor-in-Chief, noted that the results “pave the way for larger validation studies of canine olfaction as a cancer screening modality, with particular relevance for populations with limited health care resources.”

The study was selected for a commissioned editorial, where independent experts examine the work and offer their own verdict to the field. Dr Shiran Shapira and Prof Dr. Nadir Arber, the authors of ours, noted the scale and the design as much as the results, calling it “an important effort to move canine olfactory detection beyond anecdotal observations toward a more structured analytical framework.” Their conclusion – “the reproducibility of the phenomenon across multiple laboratories suggests that a real biological signal exists.”

Akash and Dr Sanjeev Kulgod, Chief Surgical Oncologist at Radon Cancer Center, were invited to the monthly JCO podcast where they discussed our work and where it leads with host Dr Davide Soldato. For a study about dogs sniffing cancer, this kind of reception from the world’s leading cancer journal was everything we could have hoped for.

The next phase
This study establishes what scientists call analytical validity, proof that the system can accurately distinguish cancer from non-cancer under controlled conditions. What it does not yet establish is clinical utility, whether deploying it in a real screening population, on people with no symptoms and no prior cancer history, produces the outcomes we believe it can.

FirstAlert, our prospective study tests the system across higher-risk but otherwise healthy individuals and people who have completed cancer treatment, to validate whether the dogs can detect a returning cancer before it shows up clinically. A larger community screening study will follow. Together they are the path from a validated study to a screening tool that works at the scale of the population that needs it.

We are enrolling for our trials. Every person who joins helps the dogs get better, the model get sharper, and the case for early detection gets stronger.

Our pack is on the scent. We hope you will join us.

A healthcare system of superlatives, creaking under strain

India is a country easy to describe in superlatives. The world’s fastest growing major economy. The most populous country globally. And the largest youthful population too, with 65% of our citizens under the age of 35, promising potential great demographic dividends for our workforce.

The trendline appears to continue when it comes to healthcare: we have the world’s largest government-funded universal health coverage scheme, the Ayushman Bharat Pradhan Mantri Jan Arogya Yojana (AB PM-JAY), which covers care for over half a billion of our most socioeconomically-disadvantaged citizens across more than 120M families. We are the biggest source of qualified doctors immigrating to the United States. And we remain a fast-growing destination for medical tourism (CAGR of 12-16% standing at $9B in 2022), with some of the lowest costs of surgery globally at competitive health outcomes.

But national potential is rarely realized in a game of averages, and these statistics obscure a deeper and more troubling truth: that our healthcare system, much like the rest of our country, is also a structure of extremes. Modernizing infrastructure and medical expertise exist in parallel with a deep and persistent inaccessibility to those resources. For the vast majority of Indians, especially beyond the borders of our metropolitan regions, the experience of healthcare is defined largely by affordability, distance, and chance.

Take, for instance, the average Indian’s ability to access primary health care: generally considered a foundational measure of an effective healthcare system, in India it is still a postcode lottery. According to an independent survey, only 10% of Indians in rural areas, where more than two-thirds of the population lives, were able to visit a primary health center within 10 kilometers of their residence, with over 90% of respondents having to travel to a different location for specialized treatment for diseases like cancer.

Secondly, out-of-pocket expenditure remains a significant chunk of our total health expenditure, with the National Health Mission’s latest National Health Accounts (2021-22) finding that 45.1% of the current healthcare expenditure is still out-of-pocket for households, disproportionately falling on rural households. This likely indicates that our health costs are rising faster than insurance can keep up, hitting a ceiling on reimbursements and scope of coverage.

Source: National Health Accounts, 2021-22

Third, we are behind the required number of healthcare workers to meet WHO standards for our population – out of the requisite 44.5 physicians, nurses, and midwives per 10,000 population, we had approximately 24 per 10,000 population in 2020 – and the majority of qualified, active health providers that we do have are highly concentrated in urban areas, leaving rural areas defenceless, where some of our most vulnerable citizens live.

When we connect these datapoints, what we see illuminated is a healthcare system cracking under stress at every level: public and private, urban, semi-rural, and rural, across socioeconomic divides except for the most affluent. Our existing models of care and financing, whether they be public or private, simply cannot scale.

It is a conundrum we cannot wriggle our way out of by pumping out medical graduates every year (and even that appears to be a question mark). The burden of care at the scale of a population like India’s is simply too massive. Moreover, we face real structural problems that continue to eat away at the foundations of our healthcare system.

Per-episode care is expensive – often prohibitively so – not just because of genuinely high clinical complexity, but also due to a growing pattern of overmedicalization. It doesn’t help that healthcare in India, like in many countries, is built around fixed-cost infrastructure such as expensive hospital buildings, medical equipment, and salaried specialists. But while this is true globally, the Indian context presents a particularly stark mismatch: fixed costs must be recovered in environments where patient volumes are unpredictable, especially in rural areas where footfall may be low, leading to a concentration of capital in urban centers, inflated margins in high-income segments, and underutilization elsewhere.

Meanwhile, because few incentives exist to prioritize long-term outcomes over short-term intervention, prevention and continuity of care remain afterthoughts. The fee-for-service model nudges providers to treat episodes, not individuals, leading to duplicated tests and gaps in follow-up. Even where digital systems exist, most are fragmented and lack interoperability. The vast majority of hospitals still run on paper, and adoption of the Ayushman Bharat Digital Mission has been lacklustre at best, with a gap between Health IDs generated and actual integration into hospital workflows.

We do not yet have a truly coherent national plan to address these faultlines, and as long as that is missing, vast chunks of our country – spanning the semi-urban and rural demographics – will remain structurally underserved, despite being where much of the country’s population and future economic potential lies. This is a failure in market design.

Put another way, we are not just short of infrastructure, we’re building the wrong kind if we continue to try to solve our volume problem by scaling up old models built on scarcity and selectivity.

Can We Skip the Unhappy Middle?

The idea that India could “leapfrog” traditional, linear models of development is not a particularly new one; in fact, we’ve done it before. One of the greatest infrastructural triumphs India has executed over the last decade may well be our leap from a largely cash-driven economy to mobile-first payments, almost wholesale skipping our credit card era. In June 2025, UPI processed ₹25 lac crores ($298 billion USD) across 19.5 billion transactions. Think about the sheer volume of capital that represents! The mind reels.

That payments system has reached corners of the country that had never interacted with formal banking, never had a checking or savings account, perhaps never even received money in any form but physical notes and coins. It didn’t just bring fintech to a country that was increasingly embracing the proliferation of smartphones at every socioeconomic stratum; it rewired our mental models of access, trust, and financial and social inclusion. The whole nation learned that with the right policy scaffolding, a system designed for interoperability, and an ecosystem approach that enables private innovation atop public infrastructure, you can democratize access at the population level.

But healthcare and payments aren’t comparable apples-to-apples. A financial transaction, even at the volume we’re seeing with UPI, can be broken down into relatively discrete, defined steps: a sender, a receiver, a sum, and a timestamp. A health interaction is anything but. It is longitudinal, uncertain, deeply personal, and intersects with many stakeholders beyond the patient and their doctor. It involves handling sensitive patient data and managing clinical risk, which in turn demands not just accuracy, but context. Consent isn’t just a legal checkbox, it has clinical implications. People don’t just trust in institutions, it is interpersonal and heavily embedded in the patient-doctor relationship. And the outcomes aren’t a complete-incomplete binary, they unravel over months, years, or even decades.

The stakes are high with money; they skyrocket when you try to design guardrails for health.

Despite this, we are nearing an inflection point coming on the back of two powerful trends that signal a shift on the horizon: first, an expansion in population-level health coverage thanks to growing insurance penetration and regulatory pushes, and second, the growing digitization of health-related processes. We believe that these twin forces create the possibility of not just reforming the Indian healthcare system incrementally or superficially, but changing the paradigm fundamentally by skipping intermediate phases altogether.

Over the past decade, India has seen insurance penetration rise from under 15% to over 40%, driven largely by government schemes like AB-PMJAY, the Central Government Health Scheme (CGHS), the Employees’ State Insurance Scheme (ESIC), and numerous state- and community-led initiatives. Meanwhile, on the private side, we’re seeing the emergence of healthtech and insurtech startups offering a menu of products such as bundled preventive care packages — look at companies like Loop Health, Plum, and Onsurity, delivering bundled health benefits through employee wellness plans for startups, SMEs, and MSMEs in a digital, largely self-serve motion.

Source: “Making quality cancer care more accessible and affordable in India”, FICCI-EY 2022

The question isn’t really whether India will be at least a partially insured nation anymore – it already is – the real question is what kind of healthcare system we are insuring people into. Are we genuinely improving outcomes, or just moving around risk in a model that is fundamentally broken?

In parallel, the digitization of care is accelerating. ABDM’s federated architecture for health identity and record-linking has at least initiated the regulatory push. In parallel, the quick-commercification of pharmacy and diagnostic platforms driven by the likes of Tata 1mg, who netted a tidy ₹2,392 crores ($285 million USD)1 in FY25, and Orange Health, looking to capture the market with their Amazon partnership, has compressed discovery, booking, and fulfillment into a single interface. Their sheer scale and reach are turning these at-home touchpoints into de facto population health infrastructure, bringing preventive and personalized care closer to the average Indian than ever before.

For the first time, the technical components of a new kind of healthcare stack – identity, data, diagnostics, claims – are starting to fall into place. The question is our ability to stitch them together into a coherent system.

Our mission, should we choose to accept it, is to leapfrog not just from stacks of paper filed in our clinicians’ cupboards to sleek digital systems, but from today’s fragmented, fee-for-service, provider-first model to a much more holistically integrated paradigm: where care is a continuum, risk can be computed and mitigated, and our organizing principle is population health. We do not need to mimic high-cost Western systems of care, nor become dependent on unnavigable payor cartels.

Lessons from Abroad

This is where other countries’ health systems offer both inspiration and caution. The U.S. model, decapitated by decades of managed care consolidation and private insurance gatekeeping, is arguably the world’s most expensive cautionary tale. Despite being world-class in scientific innovation and specialized medicine, access is still profoundly unequal. Over 30 million Americans remain uninsured, and an unplanned ER visit can involve multi-thousand-dollar bills.

Contrast this with Brazil’s Sistema Único de Saúde (SUS), a universal public health system launched in the late 1980s. The SUS covers over 200 million people and emphasizes primary care as the cornerstone of national health delivery. Through decentralized funding, community health agents, and concerted local governance, Brazil has significantly expanded access in rural and semi-urban areas. At a GDP-per-capita of $9,960, nearly three-and-a-half times that of India’s $2,880, it may not immediately seem like an instructive analogy, but there are useful lessons to be learned. When the SUS was first conceptualized at the end of the 80s, Brazil was coming out of decades of considerable political and economic turmoil, and yet from the very beginning the new health system was seeded with a drive to decentralize its governance structure, improve cross-level coordination, and include rural areas through community participation.

We can take our pick across space and time to see this hold true: consider China’s Mao-era “barefoot doctors” program in the 1960s and 70s which deployed millions of minimally-trained health workers, often farmers, to provide basic care in rural areas, dramatically improving metrics like infant mortality. Or perhaps Rwanda’s Community Health Cooperatives, which have achieved remarkable health outcomes with limited resources, connecting village health workers to digital systems for training, supply chain management, and patient tracking. Kenya’s mobile health platforms are making basic healthcare accessible via SMS and mobile payments in areas with little to no infrastructure.

The takeaway isn’t that India should copy any of these programs or models in their entirety, or even at all. Our demographic density and diversity, governance structures, and other peculiarities likely make such transplantation infeasible anyway. But these programs were often designed and set in action in constrained environments, and still achieved notable, quantifiable success in improving health metrics.

So to bring it back to where we began, yes, the analogy to UPI is useful — but only up to a point. The vision aims true: an interoperable, modular, digital-first infrastructure can create inclusion, scale, and innovation. The nature of the service, though, is radically different. Financial infrastructure moves capital. Health infrastructure must always keep the human at the centre, girded by insight, empathy, and action.

Regardless, the fundamental insight remains: good rails are a force-amplifier. If we lay the foundations right, others will naturally build on top of them, and new models of care and coverage can be layered atop an intelligent health backbone.

But that backbone must work for us.

When A Technical Path Meets Economic Need

To imagine a healthcare system that works for every Indian, we must first dismantle a deeply, and often unconsciously, embedded assumption: that quality healthcare is the domain of the affluent. For decades, health outcomes in India have mapped closely to income, education, and geography. But if India is to meet and sustain her economic growth and demographic dividends, this toxic alignment must be broken. In other words, we must actively decouple wealth from health.

India 2 and India 3, the emerging and underserved population cohorts delineated in Blume Ventures’ Indus Valley Report, represent the majority of the population. India 2, equivalent to Indonesia, is the aspirational middle: white-collar workers, gig economy participants, small-town professionals who often fall just outside the reach of government safety nets and struggle to afford decent-quality private care. India 3, equivalent to Sub-Saharan Africa, includes those in low-income, informal, or subsistence work, for whom even minor illness can result in financial catastrophe.

These groups are not outliers. They are the median Indian — and the majority.

From a business standpoint, this is a massive underserved market hiding in plain sight. India 2 and 3 represent not just the bulk of the country’s population but also the next billion users of new-age healthcare solutions — consumers who are digitally reachable, increasingly aware of their health needs, and gatekept from today’s care delivery and financing models. The conventional point of view has historically dismissed these segments as unviable due to their price sensitivity and the high distribution costs of reaching them. But software is steadily eroding those barriers. Nearly every part of the healthcare stack has a digital-native player operating in it: companies like Pristyn Care digitizing patient onboarding, Niramai and Qure.ai enabling AI-powered triage, MolBio building a point-of-care PCR platform, Sunfox/Tricog building pocket-sized ECGs for remote monitoring, and Even Health enabling regular preventive touchpoints, now allow providers and payors to reach lower-margin segments with lesser overhead and higher precision. The infrastructure to deliver care asynchronously, outside hospitals, is growing.

In this light, the challenge is no longer one of pure economics but of imagination. If you reframe healthcare as a platform business built on data, services, and recurring engagement, just like any other, then India 2 and 3 are not merely high-risk write-offs; they are a high-volume, long-lifetime-value customer base waiting to be tapped. The macro tailwinds are blowing. What remains is a model that can optimize unit economics in these segments without collapsing under the weight of legacy cost structures.

We are beginning to see the contours of this opportunity emerge. Narayana Health’s push toward becoming a “payvidor” by layering their clinical delivery with digital engagement tools and recently rolled-out insurance products shows how a provider can own more patient touchpoints while better controlling costs and outcomes. Even Health, starting from the payor end, now bundles insurance with increasing integration of care navigation and preventive check-ins to reduce claims costs while improving member experience — and apparently reducing hospital readmissions by 50%. PB Healthcare, a venture of PB Fintech that owns Policybazaar and raised a $218M “seed” from General Catalyst, is expanding into the offline world with a marketplace-to-clinic model, using data from online activity to drive real-world patient footfall.

Beyond domestic moves, we see behemoths like MunichRe and John Hancock coming together with disease detection companies like GRAIL to offer policyholders access to the Galleri blood-based MCED test. Like in the case of Even, the play signals a larger realignment towards long-term engagement versus episodic intervention on the part of payors, where early disease detection becomes a lever for long-term value through population health management.

These are not fully integrated systems yet, but they signal the direction of where the market is headed: vertical and horizontal integration that owns more of the patient journey and aligns incentives around long-term engagement rather than purely episodic intervention.

Successful models will pair this integration with digital infrastructure: identity-linked patient engagement, data-driven risk modeling, and outcome-based payment mechanisms. The economics can shift to manage the inherent complexity of healthcare. A growing body of global evidence suggests that preventive care improves health outcomes and makes healthcare more cost-effective if not cost-saving, that retention reduces churn and improves continuity of care, and that tailored risk stratification allows insurers to better price their services and control loss ratios.

We are moving towards a convergence where the technical scaffolding to manage healthcare at population scale exists and economic and demographic tailwinds demand its adoption. But infrastructure alone will not write our destiny.

The crux this will depend on is whether India can design a healthcare stack that is not only scalable but also inclusive, intelligent, and interoperable.

HealthStack 2035

If India is to chart her own path toward truly universal healthcare, we can’t just push on scaling whatever currently exists and risk amplifying existing inequities. We have to architect a digital, interoperable, and modular infrastructure that meets our citizens where they are, not just in urban hospitals but across small towns and rural villages, and ranges the full arc of the care journey.

We believe that at the base of this future system lies identity. Every patient interaction, whether with a clinic, insurer, diagnostic lab, or telemedicine platform, must link back to a singular persistent, secure identity layer as soon as the touchpoint occurs. The ABDM’s Ayushman Bharat Health Account (ABHA) offers a good foundation for this, as all that is needed is an Aadhaar card. This enables longitudinal tracking of health records across providers, creating a comprehensive view of a patient’s medical history.

Above this sits a federated data layer, FHIR-compliant and based on consent, where this patient data flows into. Today, India’s EHR landscape is fragmented: some hospitals have stumblingly adopted electronic records but most systems remain siloed and incompatible, if not stuck in the 80s with filing cabinets. A federated data model would allow patients to control their data, mitigating privacy concerns, while enabling secure and seamless information exchange across healthcare facilities. This kind of data lake doesn’t just unlock better diagnostics or continuity of care: with high-fidelity, real-time datasets, startups can build triage tools, personalized wellness pathways, and improved clinical decision support.

Layer three is the risk engine, where data works with intelligence. Moving beyond static actuarial tables, AI-powered engines trained on this consented health data can produce dynamic risk scoring. For insurers, this enables clever underwriting. For providers, it flags at-risk individuals early. For patients, it offers a better pathway to care, particularly in regions where face-to-face care is infrequent or delayed and this kind of silent, ongoing triage could be lifesaving.

The fourth layer is the payment logic. Our current fee-for-service system maximizes volume over optimizing for outcomes. Measures like bundled pricing and micro-capitation could incentivize providers to actively keep populations healthy versus concentrating treatment on the acutely sick. The National Health Stack proposed by NITI Aayog envisions something similar: a claims platform built to support automated approvals and fraud detection. But likely what the system needs is a measure like Diagnostic Related Group (DRG)-style payments, where providers are paid to treat conditions instead of charging for each service. PM-JAY already demonstrates some success here.

With this core infrastructure in place, a new generation of innovators can begin to build. Imagine: fintech-healthtech hybrids offering real-time micro-loans for procedures based on verified insurance coverage, AI-driven mental health triage platforms routing users to counselors based on linguistic and cultural context, local pharmacies evolving into care hubs with diagnostic integrations, truly personalized preventive health delivered in a subscription model and powered by dynamic risk scoring. The possibilities are immense.

These are not moonshots, for the record. We believe they are the natural extensions of what becomes possible when identity, data, and payments coalesce. But digital public infrastructure alone is not a health system: driving operational efficiency in the physical layer matters too.

This stack must extend into the real world along multiple axes, starting with asset-light care centers deployed in areas lacking health infrastructure today, through schools, Panchayats, and mobile vans, perhaps even staffed by trained community health workers who understand local context. Once equipped with point-of-care diagnostics and digital connectivity, these centers can function as nodes for data-entry and basic triage. Critically, they become tangible extensions of a virtual system, helping reach the masses not by asking them to come to care, but by bringing care to them.

High-frequency, low-touch screening is one of the most critical levers for early disease detection, particularly for diseases with high cost of care, like cancer. In many parts of India, people don’t even get diagnosed because they never get screened for lack of awareness or fear of the financial implications or simply because there is no infrastructure to do so affordably and regularly. This is where diagnostic innovation becomes essential. At Dognosis (woof!), we’re developing a non-invasive, breath-based diagnostic platform called BreathEasy that can already detect multiples cancers at early stages, including 87% at Stage-1 (preprint). Tools like ours don’t require specialized workers or complex equipment. You could administer them in remote locations, with minimal infrastructure, and plug the result directly into the health stack via the linked patient’s identity to triage escalation. This kind of frictionless screening could massively expand the catchment area for early intervention to target the most consequential and cost-heavy conditions, especially in limited-resource settings.

Lastly, this stack also enables smarter insurance design. Microinsurance, sometimes deemed unsustainable due to adverse selection or fraud risks, can become viable when underscored by predictive models. Dynamic underwriting and tailored premiums, layered over a risk model that draws on health stack data, allow payors to serve India 2 and 3 profitably, not just out of the goodness of their hearts. NITI Aayog’s report, Health Insurance for India’s Missing Middle, is a useful framework to think about how such a microinsurance scheme might be structured, keeping in mind affordability.

Source: “Health Insurance for India’s Missing Middle”, NITI Aayog 2021

We know that digitization alone is not a panacea and that technology is a necessary but not sufficient condition to achieve the kinds of gains we are looking to unlock; it can only succeed when the underlying structures of public health are solid.

But it does offer something we haven’t had before: a coherent architecture on which an inclusive healthcare system could be built.

What We Choose to Build Next

The path forward is not easy.

But it is technically viable and economically aligned with the needs of a population racing ahead of its crumbling healthcare infrastructure. It is also time-sensitive. The convergence of rising insurance coverage, increasing smartphone access, the early success of ABDM’s digital pushes, and accelerating private innovation has created a rare window of opportunity. But that window won’t last forever.

We have to remember that India has done this before: with Aadhaar, with UPI, with Jan Dhan, with FASTag. We have an unparalleled digital public infrastructure, and each of these systems succeeded not because of private genius alone, but because of public effort designed to foster private creativity. We’ve got to bring the same lessons to transforming our health system.

The question is not whether transformation is possible, but whether we choose to invest in it and bear the discomfort of its growing pains. That decision will shape not just clinical outcomes, but how people relate to the healthcare system itself: whether they trust it or fear it, engage with it continuously or avoid it until crisis, invest in staying healthy or spend reactively on getting well.

We stand at an inflection point. The demographic dividend that stands to power our economic growth depends entirely on our ability to keep that population healthy, productive, and financially secure in the face of sickness and catastrophe. Getting this right is, and we do not say this lightly, the foundation of everything else we hope to achieve as a nation.

A future of health - for everyone, everywhere - regardless of wealth, is waiting for us. So, will we build it?

Greetings from your favourite interspecies team!

The last two quarters are when Dognosis firmly bounded out of stealth with a big bark, letting the incumbents and status-quo experts know we weren’t just playing fetch - we mean business.

Wins -

We won a 75,000 USD contract from Singapore’s Home Team Science and Technology Agency (HTX) for a field-validated use-case of our canine BCI - embedding multimodal neuroAI predictions of canine-detected odors into a 2D map. The contract is part of HatchX, a program that promises deep and sustained collaboration with the Singapore govt upon success.

We’ve landed the end of our Phase-2 study evaluating our system’s ability to detect multiple types of cancers from breath collected in a face-mask. In over 1500 novel test samples, we achieved a 90%+ sensitivity and overall accuracy in detecting 6 broad groups of the most commonly occurring cancers, many of which have no current standard-of-care method of screening. Our ability to detect true cancers stayed consistent across early and late stages and is a step-change advance compared to any other test on the market.

Parts of this Phase-2 data was accepted at the 2025 Early Cancer Detection Conference, a world leading avenue organized by Cancer Research UK (CRUK), Stanford, and the Knight Cancer Center, as well as AACR’s Frontiers of Cancer Science, at the National University of Singapore. Our full publication is on its way to publication in a leading high-IF journal.

We presented at ASCO 2025, featured in the keynote address at the International Working Dog Conference, spoke at the Indian Institute of Science’s Brain, Computation and Learning workshop, and presented talks at IISc’s Center for Brain Research, Foresight Neurotech Group, and Behaviour 2025.

We’ve welcomed multiple new team members including an Indian Army special-ops veteran, a K9 specialist from the Israeli Oketz, a Neuroscience PhD with a background in cognitive decision making from a leading IIT lab, an electronics and firmware engineer with a decade of experience in R&D environments, a passionate and sharp founder’s office associate, two cracked software developers, including a 20 year old who built a neural network, in Rust, from scratch, two driven clinical research coordinators, and a cognitive neuroscience data scientist.

We were also happy to have hosted 10 amazing summer interns from some of India’s best universities who significantly accelerated short and medium-term projects. At the end of July, the Dognosis Pack was 50+ strong.

We were featured on page 3 of the Deccan Herald, South India’s leading newspaper with a print readership of over a million, Swarajya Magazine, a leading national online magazine, and Scientific American, the premium international science magazine. We also featured in an Instagram reel with over 150k views and a mini-viral tweet.

We launched a 2 minute about-us video, crossed 5k followers across socials, and had a super fun time at our Pack Retreat last month!

Parting Barks

There’s a lot of work going on under the hood, see Shivam’s, Aabha’s, and Anagha’s posts to get a peek!

We’re actively looking to hire for multiple roles, especially an engineering manager and developers - https://www.dognosis.tech/careers.

We’ll be warming up to start raising our seed round in the coming weeks. Introductions to founders who’ve raise seed+, angels and VCs most welcome!

Warmly,

Akash, Itamar, and the Dognosis Pack

Agreement on the definition of deep-tech may not yet be consensus, but a useful definition nevertheless exists — technology that enables radically new capabilities for humanity and a step-change in the possibility-space for our species.

One such step-change - the co-domestication of another species leading to a wholly unique melding of minds, where cognitive abilities of differentially embodied intelligent agents begin to act in unison, where each species evolves to possess new capacities for action in physical and social worlds.

The proto-typical example of which definitively changed the arc of two species that came to dominate Earth in the Second Millennium.

A technological process whose precise origin is still hotly contested but is generally understood to have been underway more than 11,000 years ago, well before the end of the Holocene and the dawn of the Neolithic, when sapiens sprung forward past the glacier melt into history as we know it.

Domestication pre-dated agriculture, which preceded – and indeed, was the necessary precursor to – the industrial, informational, and intelligence ages. Arguably, domestication was itself a necessary step before agriculture, the crucial domino that needed to fall before humanity would be able to raise sentience from sand a few thousand years later.

“Remove domestication from the human species, and there’s probably a couple of million of us on the planet, max,” says the archaeologist and geneticist Greger Larson. “Instead, what do we have? Seven billion people, climate change, travel, innovation and everything. Domestication has influenced the entire earth. And dogs were the first.”

Greger Larson, archaeologist and geneticist, The Atlantic

And dogs were the first.

Or at least, the wolf-like ancestor that would become the dog. A fellow apex predator, living in highly social and complex packs that were capable of impressive feats of collective intelligence, that betted on being better off with four legs on the ground, a tail, and befriending the fearsome monkeys with the fire-summoning powers.

The canids gained a powerful ally — an advanced simian possessing state-of-the-art weaponry, significant physical endurance, and sophisticated tribal intelligence.

The sapiens, on the other hand, would have been equally pleased with their side of the bargain. The proto-dog was a hunting partner with a fierce set of fangs in a 40 kilo muscle-suit, yes, but they were also much more: the two most consequential aspects being - a nocturnal sentinel network that allowed for deeper REM cycles and an olfactory sense-making system granting access to a far richer Umwelt. Both carried weight of immediate practical import.

And so, a partnership was natural, a symbiosis if you will, an inter-mammalian entanglement of monumental consequence.

Walking beside us, canis familiaris was close witness to the major eras of human history - agriculture, the domestication of livestock, the wheel, writing, paper, and the printing press, the steam engine, electricity, and the internet, and of course, artificial intelligence.

Uncannis Familiaris, A Dognosis Manifesto

The proto-dog is one of humanity’s most primal technologies, arguably fourth behind only fire, shelter, and clothing. Just as the former three have undergone technological evolution – with current frontier forms taking the shape of nuclear fusion, skyscrapers, and space-suits – the dog too has morphed too, the apex of which could be said to be the para-trooping Belgian Malinois.

Or have they? Such is only really true for a pawful of cases. The vast majority of canines today live with their humans in much the same way that they did hundreds, or even thousands, of years ago. The leash may be nylon rather than hide-skin and we may have tastier treats, but nothing fundamental has changed. Indeed, most of the dogs and humans of past eras arguably accomplished more impressive feats together more often than they do today. We have arguably regressed, rather than progressed. Witness the pitiful pug in a purse gasping for air from a nose engineered to be cute rather than functional.

But the few exceptions are nevertheless illuminating. A special ops Belgian Malinois can indeed paratroop out of a moving helicopter, and hit the ground running (relatively), triangulating traces of explosive and weapons by his nose that are invisible to the best radar. A mixed-breed dog riding in a speedboat can sniff out the smell of killer whale scat floating in the vast Pacific. A biomedical detection Labrador is capable of sniffing from a few milliliters of urine the difference between a low and a high Gleeson score of prostate cancer, a stupendous feat that cannot be replicated by any machine today.

All three dogs above display an olfactory wizardry that continues to grant us unmatched access to the world of scent. Remarkably, what was true of the proto-dog, remains true all these many technology branches later. We have not cracked the olfactory code but we can sniff it, through an uncanny yet familiar interspecies alliance. And while the periodic breathless article reports on the e-nose being close to taking the sniffing crown, the canine nose remains the top dog, on the battlefield, at the airport, and in the lab.

In many ways, dogs are one of humanity’s oldest R&D project. They are also our deepest - a technological step-change that provided humanity a multi-use platform with olfactory capabilities that we still cannot best today. For 11,000 years and counting we’ve had access to superhuman capabilities, but we’ve been restricted by a primitive interface - food and a handful of monosyllables. It's time to take the synaesthetic symbiosis that we've been honing for millennia and unleash the superpowers hiding in furry sight.

Before I could even walk, I had paws padding around me.

My parents had a white Labrador named Crystal before I was born, and when I was barely two, she gave birth to eight puppies right in our living room. There are blurry photos of me crawling into their basket, tiny hands reaching out to cuddle. That was my first “lab experience.” Fitting, really, because this summer, I ended up back in a lab, with dogs!

We’ve always had dogs around. Some were ours, some just wandered in and stayed. But Ace, our current black Lab, is different. He was found abandoned - all ribs, nerves, and mistrust. For weeks, he wouldn’t eat from our hands. I remember just sitting near his bowl, quiet, patient, waiting.

Now he’s the loudest, goofiest part of the house. He “helps” with yoga by lying across the mat, insists on barking during Zoom calls, and sighs dramatically when denied a second breakfast. He’s also the reason I care so deeply about the work I’m doing now.

But dogs are only one part of the story.

I’m in my final year at Plaksha University, studying Biological Systems Engineering, a field built at the intersection of biology, computation, and systems design. My coursework has ranged from deep learning and data science to systems biology and genetic engineering. So when I came across Dognosis – a deep-tech neuroscience startup decoding how dogs detect disease through scent – something clicked.

This summer, I joined Dognosis as a Biological Systems & Applied Research Intern. From day one, I found myself wrestling with one of the most challenging problems at the intersection of machine learning and animal science: how exactly do you teach a machine to understand canine cognition by analyzing a dog’s movement and behavior? The challenge isn’t just technical, it’s foundational: great models require great training data, and in our case, ensuring the quality of training data demands consistency in annotating dog behavior.

My main project is designing a computer vision pipeline to help detect and quantify dog behavioral cues during scent detection sessions, without relying on human cues that could introduce bias. The raw data is complex: ten Raspberry Pi cameras capture multi-angle video during each session, and the dogs are in constant motion, sniffing, pacing, turning, indicating. The pipeline combines several specialized tools that solve specific pieces of the puzzle. DeepLabCut handles pose estimation, tracking anatomical landmarks like eyes, noses, and tails across video frames with accuracy that nearly matches human labeling. But there’s still a gap between identifying a dog’s positions and being able to map them to behaviors – that’s where DeepEthogram comes in, to classify behaviors from movement and features.

Pose trajectories are then passed through B-SOID for unsupervised clustering (identifying behavior patterns without predetermined categories) and A-SOID for supervised clustering (recognizing behaviors we have defined). The goal: identify subtle behavioral “indications” when a dog signals a positive (cancer) sample, and do so with no bias from the trainers who might unconsciously influence the detection process.

It wasn’t easy. Syncing footage across the ten cameras, reducing noise, and stabilizing training were all recurring challenges. DeepEthogram, in particular, is very sensitive to frame drops and annotation inconsistencies. Eventually, the models started to learn – and more importantly, they started to generalize across sessions. We’re now working toward integrating these predictions into a larger real-time behavior prediction framework, complete with confidence scores to reflect behavioral uncertainty.

In parallel, I’m working on developing the SniffNet Learning Module, an educational annotation tool on Udemy. As I mentioned earlier, great models require great training data, but manual annotation can be time-intensive and laborious. SniffNet solves two problems in one: it is designed to teach students computer vision and how to annotate behavior, while helping to generate high-quality training data for our model. It’s a five-day interactive course with video lectures, in-browser annotation tools, and feedback loops. The idea is simple: instead of training people separately and then having them contribute annotations, make learning and contributing the same thing. SniffNet merges human-in-the-loop learning with real-world model development – a different philosophy on how scientific research can interface with education – and it’s one of the things I’m most proud of having worked on.

The scope of my internship extended beyond just computer vision into early-stage work on how we understand differences in individual canines’ detection capabilities. I’m working on an exploratory project in olfactory receptor sequencing, investigating whether gene overexpression or underexpression can lead to some dogs excelling at scent detection more than others. I conducted an independent site visit to an external lab, coordinated with their research team, and helped assess feasibility for tissue-level sequencing of the olfactory epithelium where scent detection actually happens. This work is still confidential and in its initial phase, but it lays the groundwork for bridging genetics with behavior and potentially predicting scent sensitivity at a molecular level.

As with every trail that we sniff out here at Dognosis, this research connects to our broader mission. If we are able to somehow identify genetic markers that indicate superior detection ability, we could select dogs more effectively, optimize our training, and maybe even understand the biological basis of the incredible sensitivity that makes it possible for dogs to detect disease.

Every project at Dognosis may be different, but they’re all driven by the same question: what can dogs teach us, and how do we listen?

This isn’t abstract work. EEG tools could help vets diagnose canine epilepsy and neurological issues more accurately. Behavior models might flag signs of stress, fatigue, or confusion earlier than humans could observe. And through olfaction research, Dognosis is working toward a bold goal: enabling early-stage cancer detection in humans using the most sophisticated biosensor nature has ever created – the dog’s nose.

What I’ve loved most though, beyond the science, the dogs, and the sheer sci-fi weirdness of some days, is the team. Dognosis is full of people who ask great questions – the kind that makes you pause mid-task and go, “Wait… why is that like that?” We have weekly syncs where everyone showcases whatever they’re building, stuck on, or curious about. It’s raw, collaborative, and deeply thoughtful. And as interns, we’ve found our own rhythm too. From making endless rounds of Vietnamese cold coffee to discussing everything from brainwaves to breakfast, it’s been just the right mix of rigorous work and genuine, kind community.

There is no shortage of work ahead at Dognosis, and the team is working on problems that some consider intractable. But if we can keep moving forward, we have a shot at creating something unprecedented: a window into animal consciousness that doesn’t rely on and isn’t influenced by human interpretation.

Somewhere between the science and the sniffs, I think I’ve found the kind of work I want to keep doing, work that’s technically demanding, biologically rich, and emotionally grounded.

And when I go home, Ace is always waiting at the door, head cocked, tail wagging, probably wondering if I brought snacks. He’s the reason I started caring about this.

Now, he’s also the reason I keep going.

Dogspeed,

Anagha

The first thing I ever designed was a fake newspaper ad for my stuffed animals. It had columns like “Bear of the Week” and “Socks Missing: Rabbit Suspected”. Nobody read it, of course. The rabbit was illiterate.

But I think that's where it began: this urge to make sense of things. Not in the rational, spreadsheet way, but in the “What's the story here?” kind of way. Design for me has always been a way of noticing, then gently pointing. “Hey did you see that?” “Doesn't that feel strange?” “Why is this the way it is?”

I grew up making things that didn't ask to be made. Absurd drawings. “Works of art” from found objects. Sparkly greeting cards for my parents so I could get an extra scoop of ice cream. Fragments of thoughts and feelings I couldn't file away, so I kept them in a sketchbook.

I never really set out to become a designer. I just kept noticing things and needing a place to put them. Like: why do error messages sound so passive-aggressive? Why do I spend forty minutes picking the right shade of blue for a button that nobody will consciously notice? Why does my brain reject every idea until the deadline is 3 minutes away?

For the longest time, I thought the pinnacle of design was a perfectly aligned layout. I'd zoom in 800 percent to nudge a button two pixels to the left, drink three cups of coffee, and call it a productive day. (Okay, maybe I still do.) This kind of perfectionism felt safe: measurable, controllable, something I could point to and say “I fixed that.” But it was also limiting. I was simply designing for the screen, for the aesthetic, not for the human looking at it.

Then I found myself in a meeting where someone said, with full seriousness, “So, the dog will sniff the breath sample and signal if it detects cancer.” And I thought, “Ah. Yep, makes total sense, I'm in.”

That's when I realised something had shifted. I wasn't just looking for perfect alignment anymore — I was looking for perfect translation. The challenge wasn't making things look right; it was making complex, beautiful, sometimes absurd realities feel accessible to people who needed to understand them.

So what's a communication designer doing in a room full of scientists, engineers, and dogs with better attention spans than I have? Translating. Not the molecules – I leave that to the dogs – but the meaning. Turning the deeply technical into something a person can feel and understand.

At Dognosis, we work with dogs who detect diseases by sniffing human breath. Yes, that's real. Yes, that's amazing. Yes, sometimes I forget and then remember mid-meeting and re-process the absurd beauty of it all. Our breath carries tiny molecules, VOCs, that say a lot about our bodies, information even we don't know yet. Dogs, with their incredible noses, can detect shifts in these VOCs that hint at stress, infection, even cancer. It's science, yes, but also mystery, translation, trust. Which, oddly, is a lot like design.

Here's what I've learned: in a company built around biological detection, design isn't decoration. It's infrastructure. When you're asking people to trust something as unconventional as a dog's nose for medical insights, every interaction becomes critical.

Take our investor deck, for instance. The challenge wasn't just explaining what we do. It was making investors believe it was possible. I started with a simple question: what's the first thing that comes to mind when someone says “medical detection”? Probably sterile labs, complex machinery, maybe some beeping. Definitely not a beagle named Whiskey with a particularly keen sense of smell.

The first version of our deck opened with technical specifications: VOC detection capabilities, clinical trial data, and complex graphs. It was scientifically on-point — and hard for a layperson to digest. So I re-designed the narrative arc entirely. Instead of starting with what makes us different, I started with what makes us familiar: laying the groundwork with the science of scent that everyone already trusts, like bomb-sniffing dogs at airports, drug detection, search and rescue, before gradually introducing the medical applications. What we do isn’t science fiction, but a natural extension of canine abilities we already rely on. The visual flow moved from recognisable scenarios to our specific innovation, each slide building credibility before introducing complexity, creating a bridge of understanding rather than a leap of faith.

The Instagram strategy presented a different translation challenge entirely. How do you introduce people to medical-grade dog detection without triggering their "this sounds too good to be true" alarm?

Stepping back and thinking about users, I realized that skepticism follows patterns. People usually question the science (“Surely dogs can’t really do that?”), are concerned about replacing human medical judgement (“Are you asking me to trust a dog over a doctor?”), or just can’t visualize how it’ll work (“What will this look like in practice?”). I developed a content strategy that balanced three elements to systematically address each layer of skepticism: highlighting the science (credible), the dogs (approachable), and the humans behind both (relatable).

Some showed our detection dogs in training, demonstrating their precision and focus — not cute pet videos (though we love those), but serious working animals demonstrating genuine competence. Others featured bite-sized science explanations, using everyday analogies to make our work feel less abstract. And some just celebrated the weird beauty of what we do: exploring the spectrum of human-canine interaction as enhancement and not replacement. Our work is driven by a deep desire to understand what we can unlock through science and technology, but that doesn’t stop us from leaning into our awe at what we’re creating.

Many people misunderstand what we do at Dognosis, and it’s been interesting to note the turning point that usually happens when people realise we're not replacing traditional medicine — we're offering an additional tool for early detection. The design challenge was sequencing that realization, creating a journey from doubt to understanding. My work sits at the intersection of all the other disciplines happening here. While the scientists refine detection protocols and the engineers build systems too complex for my brain to comprehend, I'm thinking about how a person first encounters this idea.

Design is, and should remain, simple. Consider our onboarding materials, like brochures for potential partners. Rather than making it dense with technical accuracy, but completely overwhelming for healthcare providers trying to understand practical implementation, I began the process by mapping the questions people actually ask versus the information we thought they needed and working from there.

Even something as simple as designing merch becomes an exercise in translation. How do you put “dogs detect cancer through breath analysis” on a tote bag? You don't. Instead, you design around the wonder of it: the partnership between human and dog, the idea that the most advanced detection technology in nature has four legs and a tail.

If you'd asked me what kind of designer I wanted to be, I wouldn't have a neat answer. I still don't. I just know I want to stay curious about things people stop noticing. To see the world as full of poetry — symbols, rituals, broken systems, tiny joys, glitchy narratives. And maybe a bit too much Helvetica. I design with a kind of layered logic: part user, part stand-up comedian who happened to go to design school.

Which brings me to why design matters in a place like this. In traditional design roles, you're often optimizing for clicks or conversions. Here, you're optimizing for comprehension and trust around something that could genuinely change how we approach early detection of deadly diseases. The stakes feel different. And the opportunity to contribute to something meaningful, not just profitable, changes how you think about things.

The impact of what we do here won’t always be measurable in traditional design metrics. Success might look like a veterinarian finally understanding how to explain our process to pet owners, or an investor asking follow-up questions instead of politely changing the subject, or a social media comment that says “I never knew dogs could do that, but now I want to learn more.”

My job isn't to simplify. It's to make complexity feel familiar, to make strangeness feel like curiosity, not fear. And while dogs don't care about typography (rude), they teach me something else: to pay attention, to respond, to sniff things out before they become obvious. Basically, to be the kind of person who doesn't wait to be told what's wrong, but listens closely enough to notice.

So this is where I'm at: I'm still the person who makes zines about existential dread and snack metaphors. Still scrolls past inspiration to look for more inspiration. Still the person who sometimes makes things that don't ask to be made.

But now, I get to channel all of that – the seeing, the noticing, the stories without neat endings – into work that might, one day, help someone catch something early. Something important. Something life-changing. We have a lot of work on the horizon. As we expand the scope of the partners we work with, we’ll need to continue communicating in ways that adapt to context, maintain scientific accuracy, and build appropriate trust.

I don't think design is about control. I think it's about care. And if I can keep designing with that in mind, with humour, with empathy, and just the right amount of unnecessary prototypes, then I think I'm doing okay.

That's the sweet spot, I think.

Dogspeed,

Aabha

Including our R&D team, our clinical team, our K9 team, our founders office, our interns, our consultants, and our dogs, the Dognosis Pack is 50+ strong.

Many noses on the scent.

Our R&D team brings rare chops of canine science, computational and cognitive neuroscience, behaviour, olfaction, computer vision, software development, embedded systems, electronics and communication engineering, MEMS, applied physics, analytical chemistry, robotics, and more. Fangs have been honed from studying/training/working at UC Berkeley, Vellore Institute of Technology, Jio, European Research Council, National Center for Biological Sciences, Defence Institute of Advanced Technology, Max Planck Institute, IIIT Hyderabad, among other crucibles of research, development, and practice.

Our clinical team is spread across 3 cities and 6 hospitals with in-depth, first-paw, and real-world experience in conducting research in the life and health sciences. From massive government hospitals, to long-running non-profit institutes, to large and up-coming Tier-1 and 2 private healthcare providers, they have seen how the everyday Indian across class and geography navigates the battle against the emperor of all maladies, and that despite the challenges and obstacles in running a multi-center n>5000 clinical trial, it can be done with a world-class rigour and adherence to the highest clinical standards, producing results that are paradigm-shifting whilst being grounded in reality of a messy world.

Our K9 team has experience working with and learning from dogs across 3 continents in extraordinarily high-stakes and high-pressure situations, such as explosives and biomedical detection, as well as incredibly sophisticated special-ops military missions.

Our founders office are building conversations and nurturing relationships with the funders, the government officials, the industry leaders, the clinicians and scientists, the regulators, and anyone else who is willing to listen about an outrageous but completely true story about how dogs can save the world by teaching the machines how to smell.

Our interns have come from India’s leading universities to spend their summer with us, including from IIT Kanpur, the Indian Institute of Science, PES University, NID Bangalore, IIT Bombay, Ashoka University, BITS Pilani, Plaksha University and Srishti Manipal. One of our interns studied EECS and Business at UC Berkeley before working for McKinsey, and another studied Computer Science at a local college after doing his schooling in a Kannada medium public school in Belgaum.

Our consultants have worked in multinational immuno-oncology companies as well as Harvard Medical School, have first-authored papers in journals like Nature and hold patents in fields such as biosensors.

Our dogs have seen the scent of cancer in over 1000 unique cocktails of breath and have learned to consistently tell this scent apart from a chaotic and dizzyingly confounded Indian buffet of controls with 90+ accuracy, with sensitivities often 95%+ in consecutive blinded sets of 100 (1:4 pos:neg). They are all passionate sniffers, the proverbial Beagle who will follow a scent off a cliff, and be sniffing all the way down. Two of them come from city shelters, pets left behind by their humans petty enough to not keep them by their side. One of them is half Indie, sibling to the millions treated as pests on the road-side. We bide our time for the moment when they prove them all wrong.

The Dognosis Pack is 50+ strong. We are young, hungry, nose to the grindstone and in a hurry to change the world, except for when we’re on our walks.

Tell your friends to howl back if they hear us on a full moon. And maybe give the next dog you meet a knowing nod while you meet their gaze.

Greetings from your favourite interspecies team!

It’s been more than a year since Dognosis was formally incorporated - we’re basically a full-grown adult in dog years - and our technology, team, and operations are a lot more mature, paws firmly planted in the ground and nose on the trail, ready to take on the world.

Wins

1. We had our first ever dataset accepted to the American Society of Clinical Oncology (ASCO), 2025 annual conference, widely considered as the most impactful avenue in oncology.

2. We had two papers from our R&D team published in peer-reviewed journals - in the Animal-Computer Interaction Annual Conference and BMC Veterinary Research.

3. We were featured on the cover of the New Scientist Christmas special and the regional section of the Times of India.

4. We have four fantastic team-members join us - a full-stack software developer with a Computational Neuroscience PhD, a sensor engineer with multiple patents/papers, a bioengineer specializing in computational volatile organic chemistry, and a founder’s office/partnerships lead with extensive experience working with state and national govts.

5. We have a great new website!

R&D

Our three core tech trees — wearable canine brain-computer interfaces, cognitive multimodal ML models, and an automated canine sniffing system — have all matured from scrappy prototypes to (almost) production-ready systems.

1. Our BCI system achieves <20kohm impedance across 8 channels, stays truly conformable during movement, and is fully modular to electrodes/amplifiers. We have a robust pipeline to go from video/scan -> NeRF/photogrammetry 3D Model -> custom helmet.

2. We’ve battle-tested the software architecture to capture 3+ real-time sensor streams (2 more on the way) with annotations from sample and trainer input, and our vision models are capable of robust predictions on dog indications, creating the scaffold for EEG models.

3. We’re a month away from achieving “Level 1 autonomy” with the detection station, where dogs can train and test on a set of 50 samples / 30 minutes, without human involvement.

We acquired IP and merged with a 4-member team that developed an early detection multi-cancer blood test using T-cell Receptor (TCR) sequencing and the body’s natural immune response. From conversations with leading researchers like Herbert Levine, we are confident in our hunch that neuroAI olfaction + TCR is a breath-blood one-two punch that is the key to solving for cancer early detection at scale. We also welcomed Dr Benny Chain, Computational Immunology Professor at UCL and world-leader in TCR sequencing, as an advisor.

We have two other exciting tech branches that are just getting off-the-ground, more to share on those soon.

Clinical

The partnerships with our 6 hospitals partners continues to grow strong and we’ve collected over 5000 unique breath simples since we started a year ago.

We are in conversations with multiple hospitals other than the above 6 about launching three new studies by June - a Phase-4 continuation of our current study of multi-cancer early detection, as well as studies in recurrence monitoring and treatment management.

We have formally begun our process to gain ISO 17025 accreditation as a biological testing lab, the only regulatory piece required to commercially sell tests to healthcare providers/payers in India, and anticipate to get it by Q3 of this year.

General

The path to real-world deployment of our test in India is crystal-clear to us - we just need to execute over the next 6 months to create the datasets, publish, present, and build trust with clinicians and hospital operators. We continue to cultivate our relationship with our two biggest champions at national healthcare chains that have 30+ hospitals across India.

We’ve had multiple conversations with veterinary neurosurgeons at leading Universities like Cornell and Texas A&M who were unambiguous about the value our EEG headset would hold for their practice.. Around 1 million dogs in the USA have epilepsy, 40% would benefit from treatment, the best treatment is curative surgery but the vast majority of these surgeries are bottlenecked by having good EEG recordings to pinpoint the location of the seizure-causing region. The same BCI system and features we built for our dogs transfers readily to veterinary applications, and working with the best vets would improve our systems by a lot, so we’re pretty excited about this direction.

We have other applications of our BCI system - “side-quests” - that we’re also hyped about exploring in the coming months.

Parting Barks

We continue to ask ourselves how we can move faster. There is the clock of runway but there is a more pressing timer. Someone in our orbit lost their dad to liver cancer last month and another had their mom diagnosed with pancreatic cancer stage-3. To hold the keys to have given them better odds but to not be there yet, and for this to be true for millions of moms, dads, daughters and sons, every year, is the real clock we are racing against. We’re grateful for all the people who were willing to back us early such that the rate-limiting factor is our ability to execute. We promise to try our best to move as fast as possible to make that not be the case.

Dogspeed,

Akash, Itamar, and The Dognosis Pack

"As man lifts his eyes to the stars and meets the other inhabitants of this planet, his horizons become widened, his science deeper, his philosophy more in tune with the universe as it exists.”

John Lilly, Communication between Man and Dolphin

Akash draws his commitment to improving healthcare back to his grandfather Yallapa Kulgod, who started one of the first hospitals in the Taluka of Ramdurg in Belagavi District, North Karnataka. Growing up in a family of doctors in the 99th most populous city in India, Akash saw first-hand how impactful the practice of medicine can be, after all, what is more radical than the transformation of sickness to health? Yet, he saw how doctors were inherently constrained by the technology they had access to, and the limited time in a day, no matter how hard his grandfather, parents, uncles and aunts worked, 24 hours was not sufficient to attend to the immense sickness in a country of 1.4B people. Thus began his search for deep leverage points - places in a complex system where interventions can be difficult but have great potential to bring about transformative change.

Itamar saw up-front how dogs are capable of stupendous feats of cognition that cannot be replicated or matched by humans, even with the best technology of the 21st century. During his time as a soldier and commander of the Israeli Defence Forces (IDF) Special Ops K9 Unit - the legendary Oketz - Itamar worked alongside some of the most elite dogs on the planet, performing explosive detection and other complex special operations that involved extraordinary challenges with life-and-death stakes. While he held immense gratitude for the opportunity to work with dogs at the highest level, Itamar did not expect to continue to work with dogs after the completion of his service, harbouring an impulse towards doing something that held greater potential for scalable impact. Yet as fate would have it, he ended up finding a path that remained entwined with canines.

Akash was fortunate to have the opportunity to study at the The University of California, Berkeley. There, he ended up taking 28 classes across 14 departments, performing research in 5 different research labs, and being involved with multiple student organizations. The common thread uniting his explorations was a fascination for the ability for cutting-edge science and technology to shine a light into the complexities of the mind, leading him to major in Cognitive Science, with a focus on Computational Neuroscience. This formed a rich intellectual substrate that he carried back with him on a gap semester in Belagavi during the second wave of the Covid pandemic.

As he observed his father managing a Covid hospital, with all the terrible trade-offs and decisions that this entailed, Akash, along with people all around the world, woke up to the fact that dogs could detect Covid-19, even in asymptomatic people, with extraordinarily high accuracies, often surpassing the gold-standard of RT-PCR. And yet, despite their performance, and the many advantages that the form factor and user interface of canine olfaction offers, dogs were not being deployed en-mass to work alongside humans to fight the viral scourge. Akash could not help but get obsessed with the question of why this was the case - where were the covid-sniffing dogs in Belagavi? The quest to find an answer to this question soon came to have a name.

Itamar was one of the 60 odd groups around the world participating in this landmark moment in the field of biomedical detection dogs - the first time humans and dogs worked together across continents to detect and monitor disease in real-world settings. As the first employee of the startup Prognose, he was initially recruited to aid the Indian Army to train dogs to stop the spread of Covid amongst soldiers at barracks in the North. But soon, he began to work on a different set of diseases that had plagued humanity for far longer than the new SARS virus. Itamar began training dogs to detect the emperor of all maladies - cancer, specifically breast cancer, the most common cancer in the world - as well as Parkinson’s - two diseases that continue to lack an effective early detection system leading to immense suffering worldwide.

Despite Itamar being well-aware of the incredible olfactory feats of dogs, he found it hard to not be skeptical of disease-sniffing dogs, sounding as it often is perceived to be, as straight out of some alternative woo pseudoscience. Yet, this skepticism began to quickly melt away as he saw the 5 Labradors on the team consistently indicating on masks that had been worn by women who had been diagnosed with breast cancer.

As it turns out, the documented history of dogs being able to detect cancer dates back as far as 1989, to a letter written in the Lancet by two London doctors. Dogs have been showed to be capable, in randomized double-blind peer-reviewed studies, to sniff out cancer with precision, sometimes even before symptoms are clinically or radiologically apparent. The reasons for their ability to remain unharnessed is largely a human one, involving solving engineering, operational, and regulatory challenges, and not the canine ability to detect disease.

6 months after graduating from UC Berkeley, and fuelled by an Emergent Ventures India grant, Akash set off to Israel for a visiting researcher stint at the Tech4Animals lab at the University of Haifa. Around the same time, Itamar was beginning to disengage from Prognose and make plans to travel in India. He had succeeded in validating the ability for the lab of labradors to sniff out breast cancer and Parkinson’s in two clinical trials, but had grown disillusioned with the positioning and business development (or lack there-of). Given the hyper-connected networks of Israel, and the nicheness of the field, it was inevitable for the two to connect in April 2023, though funnily enough, they switched countries before they first spoke to each other, as Itamar had already arrived in India.

Over a series of Zoom conversations, often including Akash’s father Dr Shashikant Kulgod, the concrete plans to startup up dognosis begin to take shape. India seemed like the obvious place to setup kennels, given the vast population, strong medical infrastructure, high talent density, low costs, and a pre-existing network. Boosted by a second Emergent Ventures India grant, this time as a team, Dognosis truly set their paws on the ground in August 2023.

The rest, as they say, is the future.

Subscribe now

Subscribe now

Chutzpah

(noun) supreme self-confidence : also nerve, gall, audacity, effrontery

from the Yiddish khutspe, from Late Hebrew ḥuṣpāh

Ten weeks in the Holy Land / Startup Nation (02022023 - 22042023)

Israel was a fitting first chapter in the dogyssey, filled with a growing sense of ground, assuredness, and clarity, with a sprinkle of za'atar and chutzpah. The tone was set within the first week with a friendly meeting with Assaf and Zvi, CTO and data scientist of SpotItEarly. Strolling in Tel Aviv’s vibrant Sarona Market, I listened to the remarkable progress in the early detection of cancer made by the intrepid crew of entrepreneurs, scientists, and labradors. Interim results from their large clinical trial found a 93% specificity and sensitivity in the detection of the four most common types of early and late stage cancers. A lab of labradors providing highly accurate, early stage, multi-purpose, inexpensive, and convenient disease screening no longer remains a pipe-dream, indeed, the future of furry dognostics is already here.

Emboldened by such happenings apaw, I set forth on my research sprint at the vibrant Tech4Animals (T4A) lab in the charming city of Haifa. Led by the laser-sharp Anna Zamansky, T4A has been developing rigorous and sophisticated machine vision models that can predict the physical and mental states of animals and thus, was a natural base for laying the foundations for deciphering the canine mind.

My primary order of business was a rigorous assessment of the canine EEG space to craft a foundational text to launch near-future research. Surprisingly, measuring the neural electrobiome of dogs has an ‘ancient’ history, with Adolf Beck recording from the canine brain 130 years ago. I read (lightly) over a 100 papers that deployed electrodes with dogs (and wolves!) and did deep dives into 22 papers that measured the EEG of domestic dogs without anaesthesia or piercing the skin. Close dissection of these studies, all within the past decade, revealed two key insights -

• we can use relatively accessible and portable neuroimaging devices to make meaningful inferences on canine cognition

• there is still a bunch of low-hanging fruit to chomp on in terms of improving hardware, data, and algorithms

Simultaneously, I fleshed out my technical and scientific know-how on acquiring, processing, and analysing EEG data — in general, enabled by an OpenBCI Ganglion and Muse 2, and for canines specifically, by cooking up pipelines and analysis frameworks for open-source and partner-shared data.1

The intangibles of Israel held as much gravity, if not more, than the concretes above. The leap of faith inherent in the independent pursuit of a lifetime project off-the-plodded path, the cracking of foreign cultural codes, the meeting with brilliant and kind people, the creation and refinement of goals, habits, and iterative feedback loops, and the two glorious sunny weeks in Tel Aviv hanging out with the amazingly life-filled Ozzy, these and more were all valuable gifts and blessings from the Holy land.

Three details that simply ask to be mentioned

• Having committed to participating in Israel’s national neurotech hackathon, I found out the day before that my collaborator and her devoted service dog would not be making the event. Left canine-less, I instead joined forces with students from the BGU to work on an EEG hyperscanning neurofeedback game. Despite a certain team member’s mournful yearning for woofs, we ended up “winning”.

• Wonderful people I met included Yossi Yovel of the BatLab and listening to plant screams fame, amazing people from Effective Altruism Israel, and a whole troop of eccentric folks at Purim, the Jewish carnival, in Jerusalem.

• I had a world-changing experience eating the best falafel ever at Frishman, TLV.

Agency

“At critical moments in time, you can raise the aspirations of other people significantly, especially when they are relatively young, simply by suggesting they do something better or more ambitious than what they might have in mind. ”

Tyler Cowen, Marginal Revolution

Eight months post graduation steeping in India (01062022 - 31012023)

The lurching lull of quiet after an eventful spring graduation week followed by a home-bound landing to the sangeeth of my brother’s wedding allowed for the brewing of what was to come. 

Emergent Ventures had been on my radar for a while and at the end of a month of mulling on what-is-life-after-college, I distilled my deepest resonances in a dream-vial of 1500 words.  Two weeks later while reading Aurobindo’s incisive writings on Nietzsche’s lurid madness in the Pondicherry Ashram, a shot of lightning brought the vial’s contents to life.

A month later, on the cusp of India’s independence day, I was in the same Udaipur hotel with a hundred incredibly talented, ambitious, and driven people, at the Emergent Ventures India Unconference. Talking with people of my generation building highly impactful, hundred people plus companies, doing sophisticated and meaningful research and journalism, and generally refusing to accept societal constraints on the realm of possibilities, completely recalibrated my notions of capability and infused me with radical notion of agency.

The next months were preparation of body, mind, and spirit and a growing realization for the need of a leap of faith towards what became a research odyssey.

Sparks

“Berkeley, where cyberpunk meets psychedelic.”

“We already knew the solutions to many of the problems we now face. What we need to do is to simply relearn how to listen.” - what a time to be alive

2018 - 2022

I ended 2021 perched on the ledge of Doe library sensing a world bristling with a psychedelic cybernetic electricity. Cognitive Science at Cal is a smorgasbord of interdisciplinarity with an overriding resonance of encounter — of carbon meeting silicon, animal meeting symbol, history meshed with future — and being at the confluence of these streams led to an overwhelming sense of wanting to build bridges.

With the doors of perception slightly ajar, dogluk began to open their eyes, poodling around at the methods of entrepreneurship bootcamp, attempting lift-off with a catalyst grant, synthesising expansive models of cognition through an honors thesis, and landing a new beginning with an optimistic vision of the future.

Names

dognosis

(noun) the art or act of identifying a disease through partnership with a dog

(noun) living together in intimate association or close union between dog and man

(noun) esoteric knowledge of spiritual truth through encounters with Dog

dogluk : nominative determinism

sympoiesis technology : the bringing-forth of the true nature of things through the making-with or becoming-with other beings.

Vision

Conceived through a light moat

her gaze that june night was a moat-light that broke through.

Distilled to a concentrate of poiesis technology, dialogical living and aligned uplift

these limitations can be transcended, not just by simply leaving behind the soil of Being, but better, by rerouting our technological prowess towards synapsing connections that burst pass the bars of our Umwelt and seed an interconnectedness that is mycelial, harmonious, and aligned from its very conception.

On-going as a work-ever-in-progress at the collaborative consortium dognosis project.

Next update will be in a few months from now. If you have read this far, I appreciate you.

Doggedly yours,

ASK

akadogluk

Subscribe now

Share

"Canis lupus familiaris, indeed; the familiar is always where the uncanny lurks." Donna Haraway, When Species Meet

"As man lifts his eyes to the stars and meets the other inhabitants of this planet, his horizons become widened, his science deeper, his philosophy more in tune with the universe as it exists.” John Lilly, Communication between Man and Dolphin

I - Poiesis Technology

Our sapien ability to manipulate both atoms and bits may be at a civilizational zenith. Yet, by being entombed in the ego-dominator culture1, the radical potential of our wizardry risks remaining unrealised and new possibilities afforded appear shackled to a system of thought centered around the Anthropocene.

Heidegger claims the essence of modern technology is an aletheia or a revealing of truth.2 He argues that this revealing has two modes, the dominant one in modern society being a ‘challenging’, that reduces all beings to vessels qualified by their degree of ‘standing-reserve’. Arguably, many trials to human and other-than-human survival and flourishing are downstream from this conception of technology.

The second mode, Heidegger calls poiesis3. Poiesis technology can be understood as the bringing-forth of the true nature of things. Poietic events “are acts of unconcealment … in which entities are allowed to show themselves.”4

II - Dialogical Living

“Animals are just like humans. We eat our fill when our gardens are full of bananas and rasa si palm fruit and they eat their fill when the fruit of the forest trees is abundant. This is their food as it is ours, for the animals we hunt are the ghosts of our ancestors transformed into game in the beginning of time” - Davi Kopenawa, Falling Sky

The core idea of poiesis pre-dates a certain fallible 20th century German philosopher. Many indigenous peoples in the Americas held the belief that humans must work dialogically to live fulfilling lives.5

Technology as a means of mastering relation is a theme resplendent in the metaphysics spun by Davi Kopenawa in his description of the Yanomami way of living.

Poiesis technology and dialogical living shine light on the pre-existing intelligence of the natural world and interweaves and amplifies this wisdom using tools we fashion from our anthropotechnics.6

"I love ChatGPT, but I'm going to bet on Mother Earth. I'm going to bet on soil." - Paul Stamets [link]

III - Collared and Un(leashed)

"Relations are constitutive; dogs and people are emergent as historical beings, as subjects and objects to each other, precisely through the verbs of their relating." Donna Haraway, When Species Meet

Humans and dogs have danced a dance of millennia. Recent studies suggest our interspecies entanglement goes back 11,000 years, sparking right at the dawn of the Holocene, when the retreat of ice allowed sapiens to spring forward and leave history in our wake. Walking besides us7, canis familiaris was close witness to the monumental eras of human history - agriculture, the domestication of livestock, the wheel, writing, paper, and the printing press, the steam engine, electricity, and the internet, and of course, artificial intelligence. And as our mastery over the physical (and informational) world grew tighter, our comradeship with canids began to contort — from convivial companions to collared captives.

The relations that dogs and humans enter into are forges that fashion the nouns of our being and the verbs of our becoming. As such, they bepaw a certain intentionality, a capacity for response. Haraway points out that such capacity “can be shaped only in and for multidirectional relationships”. If we don’t allow for response-ability, if our “material–semiotic relating breaks down or is not permitted to be born”, we consign our dogs into becoming “biological artifacts” shaped by us to serve our ends.

We can already see the barring of response-ability in the language of ‘owner and pet’ and the curious assumption underlying the necessity of asking an owner to pet a pet. The consequences of such petization are far from petty, as the cruel health effects of in-breeding make clear. Lapdogs are not laptops, their heat is not mere epiphenomenon, and the gasping of the pug from a nose engineered to be cute can be seen as a direct consequence of our broken paradigms of relating.

A pointed question at this point is if such a gulf really matters when we are surrounded by the horrific and genocidal chasms such as the meat-industrial complex that emerge from our capitulations to Moloch. Yes, we bar our dogs from living their lives to their full canine potential but aren’t there bigger fish to unfry? While it is indeed proper to acknowledge that canines fare better than most, it is precisely this closeness to the open that makes the search for ways to improve our relationship all the more vital. For if our best friends are kept collared, leashed, and gasping for air, the prognosis for homo sapiens to engender healthy and shared inter-dependence is dire indeed. We must strive to do better and we are most fit to start at home.

“The relationship between humans and dogs is the prototypical instance of intelligence alignment between beings of varying capacities.”

IV - Nosing Towards Aligned Uplift

‘To be able to break through to understand the thinking, the feeling, the doing, the talking of another species is a grand, noble achievement that will change man's view of himself and of his planet.’ - John Lilly, Communication between Man and Dolphin

“The spiders have equivalents of the Prisoners’ Dilemma, but they think in terms of intricate interconnectivity, of a world not just of sight but of constant vibration and scent. The idea of two prisoners incapable of communication would not be an acceptable status quo for them, but a problem to overcome: the Prisoners’ Dilemma as a Gordian knot, to be cut through rather than be bound by.” - Adrian Tchaikovsky, Children of Time

The dominant theme of the 21st century, assuming we have historians in the 22nd and beyond, will likely be the advent of artificial intelligent systems. The rise of silicon intelligences is simultaneously exhilarating and terrifying - a civilisational roller-coaster whose tracks are being laid out on the fly. Metaculus currently predicts Artificial General Intelligence to be ten years away, and 31% of ML researchers in 2022 (n=355) believe AGI-type systems would make the world markedly worse. The consequences of truly alien, non-carbon based intelligences, capable of immediate replication and continuous self-improvement is already vividly depicted by plenty of sci-fi creations - HAL, Skynet, Ultron - with none truly capturing the inanity of sapien capitulation in the current moment. Prompts of doom-foom rarely make for compelling manifesto material, but the year is 2023 and such is but necessary when conjuring up an alternate vision of technology and the future.

The resonance of uplifting biological intelligences is made all the more deeper when considering the alternatives. While Sam Altman is right that biology has limitations, these limitations can be transcended, not just by simply leaving behind the soil of Being, but better, by rerouting our technological prowess towards synapsing connections that burst pass the bars of our Umwelt and seed an interconnectedness that is mycelial, harmonious, and aligned from its very conception.

Timeline

The first published account of canine medical intuition dates back to 1989, where two London doctors Hywel Williams1 and Andres Pembroke relate the curious anecdote of a woman learning about her malignant tumour from the incessant pawing of her dog. From their <300 word piece titled ‘Sniffer dogs in the melanoma clinic’ published in the Lancet -

The patient first became aware of the lesion because her dog (a cross between a border collie and a doberman) would constantly sniff at it. The dog (a bitch) showed no interest in the other moles on the patient’s body but frequently spent several minutes a day sniffing intently at the lesion, even through the patient’s trousers. As a consequence the patient became increasingly suspicious. This ritual continued for several months and culminated in the dog trying to bite off the lesion when the patient wore shorts. This prompted the patient to seek further medical advice. This dog may have saved her owner’s life by prompting her to seek treatment when the lesion was still at a thin and curable stage.

The authors trot out a hypothesis that malignant tumours and their “aberrant protein synthesis emit unique odours which, though undetectable to man, are easily detected by dogs”. In other words, dogs can sniff out the stench of cancer.

Four years later, the letter happened upon a dermatologist in Florida by the name of Armand Cognetta, who had been rueing that 1 in 5 melanomas went undetected. He decided to set out to work with veteran K-9 trainer Duane Pickel and his schnauzer George to see if a dog could be trained to detect melanomas. George’s training began by searching for test tubes with melanoma samples hidden around the house, progressed to identifying the right tube amongst 10 holes in a cardboard box, and ultimately to samples hidden under bandages on a willing nurse’s body. George’s capabilities were then ad-hoc tested with a grand total of 7 patients who were suspected, but not confirmed, to have melanoma and identified cancer in four, possibly five, depending on how you go about the tricky business of evaluating what is the ground truth. Hywel and John Church2 relayed George’s exploits and another anecdote of melanoma sniffing in yet another letter to the Lancet ‘Another sniffer dog for the clinic?’ in 2001.

It was finally in 2004, 15 years after the first Lancet letter and 18 years ago, that disease detection dogs officially marked itself as a scientific field of inquiry and the ‘dog scan’ was put under the experimental scanner. A team of UK doctors investigated the bladder cancer sniffing abilities of 6 dogs and how well they did compared to chance in identifying which of 7 samples contained a cancer sample (expected value of 1/7 - 14%). After 7 months of training, the dogs achieved a mean performance of 41%, selecting the right sample on 22/54 occasions. While this wasn’t the exemplary performance one would hope for (and which would soon follow), it was striking evidence that something was indeed apaw.

Pounce of Nuance

A quick note on terminology. Canine medical intuition comes in two breeds - biomedical detection dogs and medical assistance dogs. Biomedical dogs detect the presence of disease by either directly smelling the presence of an infective agent (virus/bacteria/helminth) or more commonly, by identifying changes in the volatilome (more on that later). Medical assistance dogs learn to identify changes in either smell, behaviour, or other mysterious processes, in individuals with diabetes or epilepsy (and other rarer conditions), and warn their human guardians of an imminent hypoglycemic episode/epileptic seizure.

State-of-the-bark

Since then, canine diagnosis has been tested in at least 16 different types of diseases in humans, animals, and even trees. A handy table from a paper published in 2021 summarizes twenty papers that investigated dognosis and included sensitivity and specificity (frustratingly the table displays no discernable order) -

Dognosis saw a meteoric rise with the arrival of the pandemic times. The same paper summarizes the twenty-two papers that used double blind randomised controlled trials to assess the efficacy of dognosis for covid19 -

Something the tables don’t include that is worth pointing out is that the vast majority of studies (likely all though I didn’t check every one) had very few dogs involved - in the single-digits.

The proliferation of scientific studies has occurred alongside the rise of commercial and non-profit organizations deploying dognosis in the real world. Medical Detection Dogs in the UK is the top dog with £4M+ in funds that has ongoing studies in Covid19, prostate cancer, Parkinson’s and malaria, among others, as well as training diabetes and epilepsy alert dogs. Cancer Dogs in Canada are testing firefighters for lung cancer in over a 100 departments in the United States. Dognosis was deployed in countries other than the 13 mentioned in the Covid table, including by the Indian army, Thailand, and Rwanda. Unsurprisingly, numerous reports wagged their way into the press over the past few years - including pieces in the New York Times, WIRED, Nature and The Scientist.

Dognosis By Dog Noses

The hypothesis offered by Williams and Pembroke in 1989 that dogs are smelling out disease has since been formalized. In a way, the stench of disease is by no means a modern theory, dating back to Hippocrates in 400 BC who used the art of smell to diagnose ailments. Indeed, the miasma theory of disease, that predominated before germ theory took over, went too far and reversed the causality in claiming odours caused diseases.

Our current understanding goes something like this. Every organism undergoes cellular metabolic processes that produce molecular by-products. Some of these by-products are low molecular weight compounds that exit the body via the air, skin, blood, or liquid/solid waste products. These volatile organic compounds (VOC) are collectively called the volatilome and changes in a disease-contingent manner. With their remarkable olfactory acuity, dogs are capable of detecting these shifts and can be trained to identify them across individuals.

In the studies reviewed above, disease detection dogs generally underwent a training and a test phase that involved positive reinforcement and clicker shaping to train dogs to learn to identify target volatilomes and indicate their presence to handlers using behavioral cues such as staring, sitting, or pressing their snout against the odor container. More recent studies have moved far beyond hiding cancer test tubes around the house and use training aids such as the German ‘Detection Dog Training System’ (DDTS) by Kynoscience that automates the training process for around $17,000.

Nosey Machines

The success of dognosis of cancer prompted researchers to look for uncanine ways to detect cancer VOCs, with a 2010 Nature paper finding some success using nanosensors. More recently, an ongoing collaborative effort between MIT and Medical Detection Dogs and others, are working to develop an electronic nose (eNose) that rivals the capabilities of dognosis. In a paper published in September 2021, dognosis was leveraged to train an Artificial Neural Net on data from a gas chromatography-mass spectroscopy and microbiota analysis of urine samples from prostate cancer individuals. The results were interesting and worth diving into in a whole other piece but the punchline was -

it appears that no unique (set of) molecule(s) can be found as unambiguous biomarkers and instead it is the scent character, as opposed to the molecular composition of scent that communicates the cancer signature to canines. GC-MS produces a list of VOCs that carry the information for each experimental case but such lists themselves unfortunately do not generalize to provide a “master biomarker template” in the form of molecular species identifications and concentrations

Parting Barks

Medicine is essentially two things - diagnosis (prediction) and treatment (causal inference). 3

To me, one of the most fascinating aspects of canine medical intuition is the fundamental questions it raises about the nature of medical diagnoses.4

This is best illustrated by the the UAE study (n=3290) that performed a Bayesian latent analysis model to compare the efficacy of dognosis vs RT-PCR. They found that dognosis was actually more sensitive than the ostensible gold-standard RT-PCR, which was being used in all the other studies and real-world settings to validate the ‘ground truth’ on which to gauge the accuracy of dognosis.

Difficulties in standardization and regulation, cost of training, lack of anonymization and the infrastructure needed to care for canines are commonly cited reasons for why dognosis is impractical on a mass-scale. A big part of the challenge is the black-box nature of dognosis that provokes incredulity and the ‘can-we-really-trust-a-dog-to-tell-someone-they-have-cancer’ disbelief that forestalls wide-spread adoption.

The current popular approach appears to avoid depending directly on the carbon-based black-box dognosis and adopt silicon-based ‘we-totally-understanding-how-large-neural-nets-work’ eNoses instead. One can potentially imagine another approach that synergistically combines carbon and silicon in a manner that augments both. It remains to be actualized what this would look like. Until then, we’re stuck with good ol’ unfurry diagnoses.

[update : in Feb 2023, an Israeli startup SpotItEarly, announced the interim results from their on-going clinical trial working with dogs to screen for cancer. The data comes from over 600 control-matched samples, consisting of the four most common types of cancers - lung, breast, colon, and prostate. A team of 10 trained labradors, working in a kibbutz north of Tel Aviv, achieved a 93% specificity and 92.8% sensitivity to all cancers, with accuracy remaining stable across early/ late stages. SpotItEarly is currently seeking regulatory approval for their screening test (which importantly, as the CTO Assaf Rabinowicz told me, is different from a diagnosis) and if successful, would launch a for-profit model (which as far as I know would be the first in the world) that uses dogs to detect disease. They claim their method is 1/5th the cost of traditional non-dognostic methods, and that detection during early stages can improve odds for successful treatment from 10% to 50%.

{update : Medical Detection Dogs, a UK non-profit, has been leading the pack in researching canine medical intuition since they began in 2004. Their top-dog status was officially confirmed by the record rubber-stamper-in-chief themselves as the British organisation was awarded the Guinness world record for most diseases detected by dogs - the record breaking number being 28. List of diseases can be found on their website, and includes changes in biomarkers/status that could lead to a health emergency, so you also see nut allergies and changes in cortisol that traditional definitions might not include.

Subscribe now