Give-to-get healthcare, new marketplaces, and ground truth

This episode of Out-Of-Pocket is brought to you by…

​Camber is rebuilding the core systems behind revenue in US healthcare. We remove complexity, automate slow manual work, and give clinics a reliable way to run, grow, and focus on what truly matters, their patients.

Our team tackles the workflow and payment challenges that drain time and create friction. If you want to build real infrastructure that improves healthcare at scale, join us.​

—

Want to sponsor something in OOP? Our new media kit is out, get it here

Some healthcare stuff I’m keeping an eye on

Quick one today cause I want to eat turkey then complain that turkey is bad. Just kidding we’re a brown household - we do Indian food and one regular chicken for appearances.

A few thoughts on healthcare trends I think are interesting and worth keeping an eye out on.

“Give-to-get” healthcare 

There are a few companies popping up where the model is “get discounted access to some expensive healthcare service in exchange for contributing something yourself”. For example, Cofertility makes egg freezing free if you donate half of the eggs to parents who can’t conceive. GoodLabs gives you a lab panel if you donate blood. It’s a healthcare 69 business model, but weirdly that term hasn’t caught on.

At a high level I think this is attractive because it makes a service cheaper for people. But also because conceptually people like donating to causes and helping their fellow humans. Even if everything you’ve read online says otherwise.

I wonder if there are other areas you could extend this model to.

• Medications - If you donate unused medications, could you get something in return like reduced copays? Medicircle makes it easy for people to donate their medications and give it to people that are uninsured (disclosure: investor).
• Registry + genetic test - If you sign up for any donation registry that requires a saliva swab (e.g. bone marrow), can you get a free ancestry report? I want to find out I’m 99% Indian, but in a way that makes me feel warm and fuzzy inside.
• Specimens + research - Can you get a free specimen collection (blood, stool, etc.) if you contribute that data to a biobank? I got extra of all of it, I can assure you.
• Training data - Can you get a discounted AI MRI read if you agree to donate that data to the model’s training dataset?

There’s a lot to think through here both on the ethics side of this (structuring it so it’s not a coercion or a kickback) and on the financial side of this (the operational costs of donation have to be lower than whatever revenue you make from the other side). But there’s something interesting here.

This can also extend beyond companies and go into health policy. For example, there’s a good reddit debate on whether you should only be allowed to receive transplant organs if you’re also an organ donor yourself. Or as a middle ground, you get prioritized in transplant lists if you’re an organ donor.

I like the idea of a healthcare system where you get things if you contribute to making the healthcare system better yourself.

Quick interlude - final day to get Healthcare 101 seats

I know you’re all at home awkwardly listening to your family say something uninformed but confident about healthcare. I’ll let you get back to that, but a few quick announcements.

HEALTHCARE 101 STARTS MONDAY - I’ll teach you everything you need to know about how US healthcare works. Reimbursement, how hospitals are somehow “always broke”, how pharma patents work, etc. Sign up here.

OUR NEW SPONSORSHIP KIT IS LIVE - We do dinners, conferences, content…you name it. I’ll sell my body (mostly my mind tbh) for it. If you’re curious about ways we can work together, we should chat. We’ve already sold most of our slots for Q1 already.

In particular, if you’re a company that makes hardware used in the healthcare industry…we have lots of new things that will be up your alley.

And a reminder, we’ll come to your company to do a live workshop for your company. How to use AI practically for their jobs, custom topics on specific areas of interest you want to dive deeper on, internal hackathons for your company. Email sales@outofpocket.health and we can figure something out.

AI as a wedge into healthcare marketplaces

I’m investing in areas where AI can be used as a wedge to create new marketplaces in healthcare. I am saying this to sound important.

V1 of tech has created marketplaces that replaced human beings that used to coordinate things. Travel agents for trips, stock brokers for shares, taxi coordinators for black cars, etc. Companies did this by forcing the supply side of the equation to standardize their offerings into the marketplace (pricing, details presented, etc.).

However, this works if the buyer knows exactly what they want to buy. One reason you don’t see a lot of marketplaces in healthcare is because the demand side doesn’t know what they specifically need and the supply side offerings are too varied + they want pricing control. This does not fit into the rigidity of a self-serve marketplace format, and instead becomes a bespoke transaction each time because the needs are so different.

For example, you can see this when you need to buy insurance off the marketplace. It’s a subpar experience because:

• What I want from insurance is complicated based on my health status, income, and who I want in-network. 
• What insurance can show me is extremely limited by the fields of the marketplace. Which is why you need to click into further details that are impossible to decipher for the average person.
• The insurance needs to have data from so many bad internal systems to answer patient’s questions, so the answers are frequently not even right.
• The interface was designed by someone who’s vaguely heard of the internet but never been on it.

Instead, things become much easier if you talk to a health insurance broker or a navigator.

However I think AI can change how we think about marketplaces in healthcare. This A16Z piece does a good job talking about ways marketplaces are changing with AI. LLMs can apply expertise to better determine the specific needs of both sides AND organize unstructured data from both sides to create a match.

Consider scheduling an appointment as a patient. I don’t know what I need yet but I feel sick. A chat interface can help triage me enough to figure out the discrete needs of my upcoming appointment (labs, 30 minute visit, not urgent). An automated voice call to doctor’s offices can figure out if the practice can service those needs. Or what if that insurance navigator for the insurance marketplace was an AI call center?

I think there are lots of ways this can also play out on the B2B side, with AI specifically tackling operational pain points that can become a wedge into a marketplace. A few investments I’ve made in this theme:

• MightierAI - Filling out requests for proposals (RFPs) sucks. MightierAI is a company that basically takes a bunch of your internal unstructured excel, PDFs, SOPs, completed RFPs, etc. and automatically fills out new RFPs for you. Through that process they essentially build a profile of your business, and can connect you to public or broker RFPs that are a good match. In a future state, you can imagine taking the same info from the buyer side to understand what they’re looking for, and matching them to vendors that fit their needs. 
• DoubleBlindBio​ - For a clinical trial, a pharma company and trial site finding each other is a bit of an art and science. The sites fill out questionnaires for each trial about what they can do/patients they see, pharma companies need to match that to the needs for their trial. This is done bespoke each time and usually through a third-party. DoubleBlind is trying to make it easier for sites and sponsors to find each other by using AI to structure all the info about trials and matching it to the data it gets from the software stack/questionnaires at the trial site.
• Photon - Photon is a prescribing network that lets a doctor give their patient the prescription, and then allows the patient to choose which pharmacy they want to send it to. This allows the patient to shop around pharmacies like they would a regular marketplace. On the backend, a lot of nifty voice AI and fax AI is done to bootstrap the network and allow the prescription to be sent to whichever pharmacy the patient chooses. Plus they can get more info than normal from the pharmacy (e.g. ready time, stock status, insurance coverage) to add more details to the marketplace for patients.

Healthcare has lacked traditional marketplaces and have typically used consultants, brokers, or other types of services firms that connect buyers and sellers. I think AI has the opportunity to create marketplaces where there weren’t any, the same way that traditional software did that in other industries.

If you’re building something here, let me know. How many shares can $10 get me?

Proxy measures vs. Ground Truth

Healthcare relies on a lot of proxies to actually figure out what happened to a patient. A lot of time the data didn’t exist to measure what we actually wanted. Today, we have a ton of data being generated that’s net new and AI is useful in parsing it out to figure out what the ground truth is.

So where in healthcare can we create “ground truth” where currently a proxy is being used? Do we have an alternative today?

• EHR documentation is a proxy for what actually happened during the visit, can a scribe provide ground truth?
• Capturing random point in time biomarkers are a proxy for the ground truth of what’s going on with a patient during the course of their lives. Instead of point in time A1c capture for risk adjustment, can we use CGMs to monitor what’s happening regularly?
• Claims data is used as a proxy for how patients are getting care across the health system, can an AI care navigation company create a more ground truth map?
• Medication fill data is used as a proxy for whether the patient is actually taking the meds. Are there more direct ways we can figure out if the patient actually took it?
• Linkedin posts about 9-9-6 work culture as a proxy for a traumatic brain injury

The reality is that most proxy measures have now been codified into things like billing, standard of care, internal processes, etc. and become ossified into the regular healthcare system. But I do think there are increasingly more ways to understand what’s actually happening,

The open question is what is the business model and who’s going to pay for it? I’m still not 100% of that, but it’s obvious to me that this will be valuable. 

Conclusions + Parting Thoughts

It’s the holidays, you can stop thinking about healthcare now. Instead think about AI.

It’s become an annual tradition to show our families some of the new updates in AI stuff, and every year they get their minds blown. This year we’re showing app builders like Replit, the new image and video generation models, and training a voice model on my own voice. 

Wondering if I might actually not get invited back to the family next year. How’s your Thanksgiving going??? 

Thinkboi out,

Nikhil aka. “FSA + black friday sales cheat code” aka. “Team Pecan Pie, you know what it issss”

Twitter: ​@nikillinit​

IG: ​@outofpockethealth​

Other posts: ​outofpocket.health/posts​

{{sub-form}}

‎If you’re enjoying the newsletter, do me a solid and shoot this over to a friend or healthcare slack channel and tell them to sign up. The line between unemployment and founder of a startup is traction and whether your parents believe you have a job.