Around 8AM on Saturday, I broke bread (a fresh bagel from Pop Up Bagels) with Dr. Neil Patel, an MD/MBA Sports Neurologist & Clinical Informaticist. While lathering on a hefty amount of cream cheese, he casually said

“the last time I pulled an all nighter was residency.”

I just laughed in the moment, but then I thought about it. This is a guy who has done his time and put in the (long) hours to get the privilege of practicing medicine at the highest level. He’s so amped about what he’s building at this hackathon that he spent the entire night vibe coding with people he just met. If I had to present a single anecdote to make my case that hackathons have their mojo back, this would be it.

Why?

As I was packing a suitcase full of duck-themed swag (more on the ducks below), my wife asked me, “Why are you doing this?” To me it was obvious, but I guess to most people, hosting a duck-themed hackathon in a city you don’t live in for hackers and healers might not make any sense! Let me briefly explain.

Hackathons?

Hackathons have a special place in my heart. About 13 years ago, I was struggling to make ends meet after a failed stint at Techstars. I got a lucky break in my career when I won an a16z hackathon at their HQ in Menlo Park. It opened up a number of doors for me and gave me the confidence to pursue the career I did. Like some of the participants from this event, I also pulled an all-nighter, and even if I walked away with nothing but a handshake from the organizers, there was a special satisfaction in knowing you pushed yourself to your mental and physical limits to try and produce something worthy of demonstration. I think it’s the tech nerd version of a triathlon 🤓

That one hackathon was so transformative for me that I’m dedicated to trying my best to recreate that opportunity and magic for as many others as I can. Also, with AI, they’re fun again because you can ship an entire product in a few hours!

NYC?

All of my connections and network stem from SF, where I rooted my life while working for Doximity. Yet, during the last year or so, NYC has been the home of an overwhelming share of founders we’ve backed. Why? I’m not exactly sure... What I can say, though, is that there’s something special happening in this city and I’m determined to be an active member of this health tech ecosystem.

Ducks?

Back in the day when you actually had to write code, you would get stumped (comically by a missing semicolon) and often waste hours (or days) trying to debug software. A fun solution when you hit this wall would be to place a rubber duck (yes, the kind Jeep owners have on their dash) on your computer and explain the problem. Magically, the solution would be telepathically transmitted from the duck’s mystical brain to yours and you’d know what to do! In reality, the idea is that by talking it out, you backtrack the facts in a different way and usually uncover paths you may have been missing. It’s the best way programmers like me had to access the subconscious AI (Cerebrum Code) in our brains to defeat difficult problems. If you ever hear someone say “thanks for letting me rubber duck you,” now you know what I’m talking about 😁

Hackers and healers?

During my time at Doximity, we were able to accomplish some very special things. There are many reasons, including a lot of luck, for things going the right way. One of the things I personally believe tipped the balance for us is the dedication we had to unifying clinicians and builders to work together in solving tough problems. Docs and dorks is one way it was phrased internally. Hackers and healers just sounds a little cooler to me though 😎 The point is, I believe strongly that when you get technical builders and credentialed clinicians in the same room, the combined brain power and unique domain expertise sparks unexpected (and sometimes remarkable) outcomes. I live for the thrill of those sparks and I hope to help empower as many people as possible to ignite them.

DISCLOSURE: I’m a little biased. My wife is a clinician and I’m definitely a hacker. We make a good pairing and I think there should be more hacker/healer partnerships in the world 🧑‍💻❤️👩‍⚕️

The trial run

This hackathon is actually the second Hackers and Healers Hackathon. I ran an incognito test in SF in June on a shoestring budget out of my own pocket and was blown away by the results. People hacking all night, boxes and boxes of pizza/caffeine disappearing, and fully formed products being demoed in less than 24 hours. A company was actually born during that event and they went on to raise money! The participants did all of this for nothing more than free food and their desire to build with like-minded people. And a promise of winning one of these shiny, gold ducks 🏆

So, on October 2nd, we ran it back. This time, I called up my friends in NYC for some additional firepower to do it right.

HUGE thanks to

• Redesign Health, healthcare VC

• Photon Health, modern prescription network

• Visualize AI, AI lab building foundation models for human perception

• TechNovaTime, deploying digital employees that work across healthcare systems

An even bigger thanks to all the hackers and healers who gave it their all for 24 hours and made this event a blast!

TL;DR takeaways

• I’m in awe of how hard people worked on these projects

  • at least 2 teams cranked through the entire night!!!

• Healers were outnumbered in headcount but made up for it in how much enthusiasm they showed

  • the vibe I got from them is it’s “addicting” to see an idea morph into a product in just a few hours

• Most of the ideas we saw were NOT viable business ideas

  • ... and that’s OK! Sometimes you hack just to see what’s possible and enjoy the journey of working with new ppl

• When you force teams to present without slides (only live product demos), you get some interesting outcomes

  • we had someone do a flip to fake a concussion mid-demo 🤕

  • there were a few demo glitches and it made the presentations feel more genuine

Tele-Vision: The physical exam via telehealth

🏆 Winner of People’s Choice

Team

• Jeremy Fryer-Biggs, MD

• Neil Patel, MD

• Eswanth Sriram Chengalasetty

• Aishwarya Patil

• Liyan Cheung

• Stefán Árni Arnarsson

Telehealth went from 10 million video visits a year to 126 million, but the physical exam never made the jump. Using nothing but the patient’s phone camera, the team ran a live balance test (three different stances) and a near-point convergence check, the kind of exam you’d normally only get in person, like after a concussion.

After the visit, Tele-Vision renders a 3D mesh of the patient’s body in the cloud, plus a heatmap of their center of gravity, so the clinician gets real numbers instead of “looks fine to me.” The team’s thesis is that within 10 years, every telehealth visit will come with a quantitative data layer, and value-based care will rely on it.

This was one of my favorite demos. The team definitely brought some showmanship, with one of them doing a flip to fake a concussion (nearly breaking a flower pot in the process, mind you). The demo also captured what’s possible with the everyday tech we already have at home, which is our phones. It was amazing to see real-time visuals that help the clinician, plus simple tests the patient can do to replicate an in-office exam. Of course, this is all theoretical, and there’s no substitute for seeing a patient in person. Still, my imaginative juices started flowing when I saw traces of the body show up live during the televisit.

LinkedIn Posts

• https://lnkd.in/p/gNB6Vhxc

• https://lnkd.in/p/g3UY68UF

• https://lnkd.in/p/g8J5_pZ7

• https://lnkd.in/p/gpujnwZZ

Congestion Watch: Catching heart failure before the hospital does

🏆 Winner of Best Use of Visualize AI

Team (this group didn’t sleep and built 2 different products 🤯)

• Jeremy Fryer-Biggs, MD

• Neil Patel, MD

• Eswanth Sriram Chengalasetty

• Aishwarya Patil

• Liyan Cheung

• Stefán Árni Arnarsson

More than 7.5 million Americans live with congestive heart failure, adding up to over a million hospital admissions and $32B a year. A lot of those admissions are preventable. In the 30 days before one, patients tend to gain weight, and putting on 2 to 5 pounds in that last week roughly triples the odds of landing in the hospital. The problem is that a bathroom scale can’t tell water weight from five pounds of pizza.

Congestion Watch uses Visualize AI’s SDK to fill that gap. You log your weight, do a quick body scan with your phone, and the app builds a 3D model of you, all on the device so your data never leaves your phone. It then compares today’s scan with earlier ones. In the demo, weight and belly size were up while muscle was unchanged, which is a sign of fluid buildup. From there, the app walks you through a symptom checklist, suggests questions to ask your doctor, and lets you share it all with your care team. Next on the list is spotting swelling in the legs and feet.

Having two clinicians on this team really showed in the problem they chose to go after. When you think of body fat percentage, your first thought is probably aesthetics, but this group found a neat way to flag a serious health problem using a check that’s normally geared toward bodybuilders and fitness enthusiasts. The product definitely isn’t ready to use today, but in terms of commercial viability, I think at-home scanning for body fat changes is worth exploring.

Volution: One orchestration layer for AI diagnostic tools

🏆 Winner of Best Use of Photon Health

Team

• Dexter Fox Allen

• Rudra Patel

• Kervens Auguste

Volution calls itself a one-stop shop for the doctor’s office. Clinics are already juggling a pile of AI tools (an eye scanner here, a body tracker there, an ambient scribe on top), and Volution sits above all of them as one orchestration layer the doctor uses with the patient in real time. For the hackathon, the team went deep on the hypertension use case and built a database that weighs a patient’s comorbidities and background before recommending a medication.

The demo was a little skit with a pregnant patient (an edge case on purpose) since so many blood pressure meds are off the table during pregnancy. The tool helped steer towards the right choice and the team confidently asked all the healers in the room if anyone disagreed. Nobody did, so we can maybe take that as a sign it worked? 😅 Once the Rx was confirmed, Photon took over the prior authorization and prefilled it right in the app, leaving no paperwork or letter writing.

These guys also put a decent amount of energy into their demo skit, which livened things up a bit. On the surface, this is not the most novel solution. A number of the ambient scribes and clinical decision-making support solutions promise to offer this sort of orchestration to organize everything that’s happening during the visit.

Why I think this demo stood out to the Photon Health judge is their decision to connect with the Photon API early, at the point of evaluating the Rx, and integrate it all the way through to really try and reduce the friction for the doctor. It was a fun demo because, in a hyper-targeted use case, you could see how patients might start to expect clinicians to have real-time AI second opinions and how it can actually build patient trust when done effectively.

Peak State: Recall your best state of mind, on demand

🏆 Winner of Most Creative

Team

• Cameron Cubra (co-founder of TechNovaTime, who was inspired to participate too)

• Molly Yang

• Vikas Rawat

Peak State helps you get back to your best state of mind on demand. During onboarding, it walks you through a memory of a time you felt a certain way and maps how you experienced it (what you saw, what you heard, what you felt) into a personal “strategy” you can save and replay later. The logic here is that simply saying your internal thoughts out loud has an effect on how you feel in the moment. In the demo, the volunteer traced back to a moment when they were hearing the organ play at Mass, loud and coming from all around.

The end goal is to trigger that strategy automatically when a wearable (like an Oura ring) picks up a certain biosignal. Think of it like a mood ring that actually provides an intervention in real time. They ran out of runway on that piece because the Oura API wouldn’t hand over an access token (a very real hackathon problem), so they swapped in a stimulus to show the concept.

This one just felt like a practical idea that everybody could relate to. We’ve all had moments where we’re feeling emotionally unregulated or trying to get to another state of mind, and this group showed a pathway to get there that seemed pretty effortless.

Why I think it stood out from a creativity perspective is that they chose not to use any of the APIs or SDKs. They brought a novel idea that resonated with most people, and then they had the vulnerability to demo a real use case that didn’t feel contrived or artificial.

The health aspect, which they teased but weren’t able to build, seemed pretty useful. I’d love for my Oura ring to tell Peak State when my heart rate is rising because I’m angry about my insurance saying they won’t pay for something, and have Peak State provide a gentle intervention to help me calm down 🧘

LinkedIn Posts

• https://lnkd.in/p/gjSEWkxr

Omni OS: Every wearable, one care manager view

🏆 Runner-up for Best Use of Photon Health

🏆 Runner-up for Best Use of Visualize AI

Team

• Vatsal Trivedi

• Orijeet Mukherjee

• Tauhid Mahmud, MD

• Mohammod Arafat, MD

Omni OS turns wearable data into something a care team can actually act on. One of the team members, a family physician, framed it around his grandma. She’s around 70, has hypertension, and checks her blood pressure by hand three times a week. Home devices can send that straight to the doctor, yet it almost never lands in front of the right person at the right time.

Omni OS pulls data from devices like Oura, Whoop, Fitbit and Omron into one view (plus a 3D body composition scan from Visualize AI’s API), where care managers can spot a worrying trend and flag it. The doctor then sees curated notes inside the EMR they already use, with a Photon-powered prescription suggestion they can tweak before it goes out. The business case is value-based care. Catch grandma’s blood pressure trending the wrong way before she ends up in the hospital, and the health system earns a quality bonus too.

There’s always a team that figures out how to use all of the suggested tools at a hackathon, giving themselves the best chance to win as many prizes as possible. This group did exactly that, so perfect score on hackathon strategy 💯

Thinking about their product in particular, as an investor, I’ve been seeing this type of pitch come through my inbox quite a bit. Some focus on a specific specialty and some try to go broad. The idea of collecting all the data from our wearables and doing something with it is an obvious one for people to tackle, and we’ll definitely be seeing good solutions come to market over the next few years.

What I liked most about this team was their very clinician-first perspective on the problem. A lot of the solutions I’ve seen lack that perspective and often miss what the clinician cares about, like minimizing noise. This group had a pulse on what the clinician wants/needs to see, and it showed in their demo.

LinkedIn Posts

• https://lnkd.in/p/gQQvmR3w

• https://lnkd.in/p/gi34QtnA

• https://lnkd.in/p/gVHYtAdJ

Milo: An AI coach for diet and training, built on body-scan data

🏆 Runner-up for Best Use of Visualize AI

Team

• Agam Gambhir

• Khalid Itani

Milo’s pitch is that your fitness regimen should be on autopilot. We already have tons of data about our bodies and there’s way too much friction between that data and the small, consistent actions that actually change things. The team name comes from the Greek story of Milo, who carried a calf every day, building strength little by little, one day realizing he was strong enough to carry a fully grown bull.

In the demo, a phone body scan captured biomarkers, which Milo combines with your weight, height, BMI and even your schedule (e.g., only free to work out Mon, Tue, Thu and Fri). From there, an AI coach builds your diet and workout plan, logs meals from plain-English messages (”this is my snack”), and tracks your workouts, all in one app instead of five. It’s very similar to Google Health Coach if you’ve ever used that.

The main difference between Google’s offering and what the Milo team built was the integration of Visualize AI’s SDK, which let you take images of yourself and estimate your body fat percentage, much like a DEXA scan. That’s a piece of information Google doesn’t have readily available or integrated into its health offering, and it unlocks a big piece of the puzzle of improving your overall health. They didn’t win the prize for best use of Visualize AI, but I thought their solution was definitely worthy of a runner-up mention. It also goes on my list of things I’d probably be willing to use tomorrow.

Off The Charts: Patient experience as a clinical outcome

Team

• Rex Belgarde, MD

• Shweta Srivastava

Off The Charts starts from a simple idea that patient experience is part of the outcome, not an afterthought. Shweta told her own story about how she went in for a CT scan in a lot of pain with no idea what to expect, was told on the spot that it would be excruciating, and spent the next 40 minutes crying uncontrollably with no way to tell anyone what she needed. A horrible experience for both clinician and patient. Had she been warned, she said she could have brought a friend for emotional support.

Rex (a doctor turned clinical product manager) walked through a solution that reads the EHR to build a persona for each patient, like how they best receive information, then layers on modules. One module rewrites patient messages in a trauma-informed way, another gives clinicians a brief on the non-medical stuff that could affect care, and embedded videos set expectations before the visit and explain the aftercare. Everything links back to the evidence in the chart for both sides.

We’ve all had those really bad visits to the doctor’s office where a little useful information up front would likely have changed how the experience went. Most practices and healthcare organizations do provide this info. Unfortunately, it’s from their antiquated portals built a decade ago. They force us to dig for the information we need and dull our brains so much with irrelevant info that it’s not worth engaging with. I’d personally rather be surprised than create a login just to read a PDF explaining my visit.

What I like about this team is that it’s a perfect example of the hacker and the healer each bringing their perspective to the solution. This fresh take on pre- and post-visit engagement for both the clinician and the patient should be standard in a few years IMO.

LinkedIn Posts

• https://lnkd.in/p/g5H8KGPW

Crosswalk: Matching employees to the right ICHRA plan

Team

• Nihir Patel, MD

• Adam Behun

Crosswalk tackles ICHRA plans, a newer setup where instead of handing you a group plan, your employer gives you an allowance and you pick your own plan on the marketplace. The team stated it’s taking off. There are about a million lives on ICHRA now, up from half a million at the start of the year, and large employers are growing almost 200% year over year. That means a lot more people are about to pick a health plan on their own. Most of us pick on premium, and the team’s position is that’s exactly how you end up overpaying.

In the demo, a sample patient named Kelly (52, high blood pressure) connected her health records, claims and pharmacy bills, and Crosswalk ranked 119 real Texas plans against her actual history. It showed which plans keep her doctors in network and cover her meds without a prior auth. On the employer side, Crosswalk uses claims data to figure out which groups should get which plans, and the employer keeps whatever’s left of the allowance. For their made-up company use case, that came out to about $1.86M in annualized savings.

ICHRA plans are definitely the hot new thing, and whether you agree or disagree with the approach, more and more people are going to use them over the coming years. There are solutions in the market, like Thatch, focused on exactly this problem, so it’s not a new business idea. What’s fun is that the team slapped together a very basic version of Thatch in just 24 hours and showed how powerful it could be. It was also cool to see a team with a healer on it attack a non-clinical problem.

LinkedIn Posts

• https://lnkd.in/p/gvxApJCD

Longevity Baseline: Catching fall risk before the fall

Team

• Frank Yu

• Guoqiang Pan

• Jane Yoo, MD

The idea for Longevity Baseline was motivated by a personal story for one of the members. Two weeks before the hackathon, Jane’s mom fell on her way to line dancing class. More than 1 in 4 adults over 65 fall every year, and these falls are catastrophic, costing the US healthcare system about $80B a year. How you walk, balance and get out of a chair predicts how long you’ll live independently, yet those measures get checked once a year, if that.

Longevity Baseline turns a camera and a lab into a baseline for healthy aging. The demo patient, a 74-year-old, scored far below average on the chair stand test and had a high lipid panel, high A1c and low vitamin D. When the doctor went to prescribe, Photon flagged a sulfa allergy and major drug interactions, with the option to send a cheaper alternative to the patient’s preferred pharmacy. It all lives in a long-term record that both the patient and provider can see.

It’s usually the non-clinicians that come in with the personal story, but in this case, it was a clinician. I enjoy seeing the perspective of docs dealing with a problem completely outside of their specialty, like a highly specialized Mohs surgeon thinking about fall risk.

There are tons of solutions in the market for managing fall risk, including the one I grew up seeing on TV ads, which gave rise to our favorite “I’ve fallen and I can’t get up” meme. The memes may be funny, but falls are not... My own mother fell off a ladder a few years ago. Luckily she avoided major injury, but it got me thinking about how dangerous something as silly as falling is. What I’d love to see is somebody giving away a free solution because we can now build and maintain it for very cheap. Without worrying about a business model that depends on selling to employers, payors, or hospitals, I wonder if someone can make fall risk detection accessible to the masses.

LinkedIn Posts

• https://lnkd.in/p/gmr_eX3H

• https://lnkd.in/p/gEZ8kuaN

Parthia Health: One patient, one passport, one complete picture

Team

• Ivan Santiago Enriquez Benavides

• Om Ajay Thakur

• Heather Song

If you have a chronic disease, your health data is scattered across hospital records, pharmacies, labs, home devices and your own memory. That makes it hard to understand your own health, and just as hard for your doctors to see the whole person. Parthia Health pulls all of it into one app the patient owns. For the first prototype, the team started with heart failure, where scattered data does real harm. More than 6 million Americans live with it, and most are juggling a pile of meds while tracking weight, blood pressure and symptoms at home, numbers that rarely reach the doctor in time.

The demo patient takes nine medications, with records spread across Epic and Oracle. Behind the app, a team of AI agents (each one named after an organizer ❤️) talks through her nine flags and explains each one in plain language. It turns what they find into questions for her care team and a one-page PDF to bring to the visit, with her conditions, allergies, meds and recent labs. Once she approves, her clinician gets the same picture plus a medication reconciliation, where Photon flags drug-to-drug interactions and allergies and suggests alternatives. In the demo, it caught two drugs that didn’t play well together.

Aggregating medical records and learning from that combined dataset is another very obvious application of AI-era software, and there have been lots of commercial attempts at it so far. The piece that was most interesting to me here was leaning on the Photon API to suggest alternatives when it spots a bad drug interaction.

LinkedIn Posts

• https://lnkd.in/p/ghTgEuq7

Nudge Lab: Matching patients to clinical trials, straight from the EHR

Team

• Nalani Kopp

• Sahil Pardasani

• Narges Razavian

• Pavel Overtchouk, MD

Nudge Lab wants to put your health data to work. Most of it sits in Epic, so the team built a three-click login that pulls your records into their platform and organizes them automatically.

The headline feature is clinical trials. Nudge Lab uses its own machine learning to match your data against trials you might qualify for, then reaches out and negotiates on your behalf. On the other side, the trial’s principal investigator gets a heads-up that there may be patients who fit, plus what’s missing for each one, so they can decide if they want the connection.

If I had a nickel for every clinical trial matching solution I’ve seen...

OK, so this is a tough idea because everyone is thinking about it... BUT this team also happens to be made up of

• MIT-trained AI engineer who also happens to be trained in cardiology

• NYU health AI professor

• AI forward-deployed engineer

• AI researcher who recently co-authored research about “investigating identity-aware bias in LLM-as-a-Judge systems through large-scale benchmarking and statistical analysis across multiple frontier language models.”

Long way of me saying, this group was STACKED with relevant talent and if they want to pursue a very difficult problem, we should let them 😂

LinkedIn Posts

• https://lnkd.in/p/gq8WECSQ

Semel: Patient-led research for rare and chronic disease

Team

• Chirag Das Manyapu

• Roshan Sharma

About 200 million people in the US have a chronic disease and 50 million have a rare one. Semel is a platform to help these people drive their own research. It grew out of one team member’s own experience of living with a rare disease.

Patients upload their diagnostics, fill out a symptom profile, and get pointed toward rare tests their own doctors may not know about. For the demo, a team member uploaded his own genome and connected it to genomics databases (including ClinVar) to generate hypotheses about which variants might be making him sick. The bigger idea is that patients with similar genes and symptoms can form their own research cohorts, then bring that organized data to researchers so a clinical trial is easier to start.

First, this team was grinding so hard they forgot to charge the MacBook they were demoing on. It died mid-presentation 😂 After giving them a moment to recharge, it was great to hear about someone truly scratching their own itch for a problem that has plagued them their entire life.

Since the inception of the internet, there have always been small groups that form around niche ideas, very unique interests or, in this case, a very strong motivation to solve a particular health problem. Reddit and other forums are thriving with groups like this.

What those platforms are missing is a data and clinical element that lets people share what they’re doing in a more objective way. If you can solve for safety, make sure people act ethically, and add very big disclaimers that what you hear from others isn’t medical advice and should be discussed with your doctor, you can unlock the power of very small but highly focused groups working to solve the problems that unfortunately don’t have a business case big enough for anyone else to pursue.

LinkedIn Posts

• https://lnkd.in/p/gw2phaCA

What’s next

This event was a ton of fun! I also learned a lot about how it can be better. We’re going to make some tweaks and likely run it back in NYC sometime in 2027. If you’re into this type of thing, give me a shout 🙌