Some crypto x healthcare ideas

Decentralized EMRs, Insurance DAOs, and Drug Picking Models

Featured Jobs

Finance Associate - Spark Advisors

  • Spark Advisors helps seniors enroll in Medicare and understand their benefits by monitoring coverage, figuring out the right benefits, and deal with insurance issues. They're hiring a finance associate.

Data Engineer - firsthand

  • firsthand is building technology and services to dramatically change the lives of those with serious mental illness who have fallen through the gaps in the safety net. They are hiring a data engineer to build first of its kind infrastructure to empower their peer-led care team.

Data Scientist - J2 Health

  • J2 Health brings together best in class data and purpose built software to enable healthcare organizations to optimize provider network performance. They're hiring a data scientist.

Looking for a job in health tech? Check out the other awesome healthcare jobs on the job board + give your preferences to get alerted to new postings.

Check Out The Job Board

Looking to hire the best talent in healthcare? Check out the OOP Talent Collective - where vetted candidates are looking for their next gig. Learn more here or check it out yourself.

---

Swallowing the crypto pill

On a long enough timeline, every newsletter eventually writes a post about crypto or has an ad about Masterworks. I guess people call this “Web 3” now but I don’t really understand the difference and at this point I’m too afraid to ask.

After seeing people around me getting hilariously rich selling jpegs while I worked hard to write newsletters like a moron, I decided it’s probably worth trying to learn a bit more about blockchain, crypto, and the projects spinning up in the space. This is a very new area for me, but I’m a big believer in writing things out as I learn about them, in the hope that if I’m wrong, then people will teach me.

There are lots of explainers on what blockchain technology actually is, the point of cryptocurrencies, etc. so I won’t go too deep into it here. Instead, I’ll jump straight into a few use cases I think could be interesting for the technology.

A decentralized EMR

Wow dude so creative, never heard that one before. This is by far the most common pitch I hear, so let’s start with the basic idea.

If you think of blockchains as distributed databases where no single entity controls it, it sounds very appealing vs. the oligopoly of EMR providers who currently get to determine all the rules of where the data goes.

This Chris Dixon post about the relationship between platforms and users solidified this a bit more for me. His point is that platforms have a predictable lifecycle:

  • They start by attracting users + third-party developers who can build features on top to make the user experience more valuable. 
  • This ends up attracting more users, which attracts more third-party developers, etc. 
  • However eventually it makes more sense for the platforms to extract increasingly more value from the developers since they have all the users now.

He gives examples like Google doing this against Yelp in search, Apple doing this by charging 30% tax in the app store, Twitter shutting off their API to third-party clients, etc. This has made developers wary of building large companies solely on the APIs of other companies.


Healthcare has really never even had real platforms. The rules around getting into the app stores of the large EMRs like Epic App Orchard etc. are extremely unfriendly to companies wanting to build on top of patient data + it took a ton of legislation to force hospitals/EMRs to create a data standard and make it easy for third-party applications to develop on top of the EMRs (and that’ll take a while to roll out). 

The fact that patients can’t move their records around easily and in a standard format has severely hampered the entire ecosystem. It means that companies have different copies of your health record with different pieces of information + choose what they can do with your data (e.g. decide which third-parties get to use it, sell de-identified versions of it, etc.). As a result, companies would rather build walled gardens of data vs. play nice with other companies just to make sure they own as much of the patient data as possible. I think contract arguments around data sharing rights have started at least 1 or 2 fistfights at HLTH.

A beautiful world is one in which I own all of my health data in my wallet and can give access to anyone that wants to see it, do something with it, and then add additional data to it. Companies CAN’T use data hoarding as a strategy because they don’t own the data, I do. 

I got a glimpse of what this might look like when the Loot project was taking off. The general gist was:

  • You could connect your crypto wallet which would then generate a textbox which included a unique character + items that only you had (aka your adventurer gear).
  • You could then connect that wallet to any third-parties that want to build on top of that adventurer gear.
  • Lots of people then built applications to bring the textbox of adventurer gear to life. Some people created applications that generated what the gear might look like, auction houses were created where you could separate your gear out and trade them, guilds like Divine Roles were created where you needed a specific piece of gear to join, some people started creating text games that you could bring your character into, etc.
  • The important part here is that you could take your wallet wherever you wanted - it was totally in control and there was no “head” that was controlling development. It was entirely third-parties developing full ecosystems around this one, standardized text box. It was also crazy to see how fast this development happened - the project dropped on 8/27 and by 9/1 there were full ecosystems.



If you squint a little, this is a model for how a health record should work.

  • My health data is in a standard format and stored in my wallet.
  • Any third-party can connect to that data, build on top of it/deliver services using it, and can add to it. The data added goes to the record in your wallet, not the record that the company owns.
  • No company “owns” the data. You do -  in your wallet. And you can give access to whoever wants it. As data gets added to your record, a longitudinal view of anyone that’s accessed or added to your data is in one place.

It’s clear there’s going to be a million challenges to something like this starting. How do you deal with data privacy if a copy of your health data is theoretically kept in multiple places? How do you set up incentives for people to host copies of the data + validate transactions are legit (aka. what crypto miners do)? How would this play with existing EMRs (there probably has to be some backwards compatibility here)?

I don’t have answers to these questions, nor do I know when a company like this would ever succeed. I’ve seen lots of attempts at this- especially in 2017 when crypto was in full swing (like MedicalChain, Patientory, etc.)- but have not seen any real adoption for any of them.

That being said, I think third-parties being able to build on top of health records is extremely important and a system like this means it’s more likely companies would be willing to build on top of health data without fear of being shut out of an API or suddenly being charged tons of fees. Patients would get to choose where their data goes, instead of health record companies.

Insurance DAOs

I’m interested in the idea of Decentralized Autonomous Organizations that provide some sort of insurance coverage. We talked a bit about DAOs in the previous post, but the general gist is that they’re organizations which have rules around who’s included/excluded and where decisions are made by the members in some form of governance with weighted voting. This idea around weighted governance/voting would make sense in an insurance pool where everyone’s actions are connected financially.

Currently, a wave of decentralized insurance products are being built like Insurace or NexusMutual. Most of these projects are creating insurance products to protect against hacks or exploits when using other cryptocurrency products. You can look through the NexusMutual whitepaper to see how this theoretically works, but here are a few key pieces:

  • The company needs “minimum capital requirements” aka the minimum cash you need relative to claims you theoretically cover. This is something every insurance company needs and the amount is usually based on the company size + underlying risk of whatever you’re covering (health insurance has risk-based capital).  NexusMutual’s model theoretically does this by issuing tokens that increase in price when there’s a lot of capital relative to claims, and decrease if the capital pool falls, theoretically incentivizing investment when the capital pool dries up. Underneath a certain ratio, it stops taking on new coverage until it meets the minimum capital requirement.
  • Claims assessment seems to be handled in one of two ways. Either the DAO uses an “oracle”, which is essentially a third-party that gets data from the “real-world” that everyone in the pool agrees is trusted. Or, votes go to the members for a vote, and people with more tokens get more voting power. There seem to be some rules around minimum voting power needed relative to the claim amount to have the transaction go through.
  • In theory, the pool gets re-invested in other liquid assets while it’s sitting idle so that it’s generating some form of return. I find it pretty funny that NexusMutual has chosen to just keep it all in ETH or other ethereum-based tokens since they seem pretty correlated price-wise to a large scale cryptocurrency hack but what do I know.
  • All the other functions are coordinated to members, third-parties, or automated and given token rewards for doing these tasks. The idea is that with more tokens at stake in this pool, there is less incentive to be a bad actor since it’ll tank your share of tokens as well. However…it also means that there’s a bit of a disincentive to pay out claims. This is the tension all insurers face, and is why it’s important to have an agreed upon + enforceable standard for payouts.


There’s a lot more in the whitepaper, but hopefully you get the gist. I think there’s probably a way to bring some of these principles to insurance products in healthcare. This would more likely than not be implemented as non-ACA compliant plans.

Health sharing ministries would be one of the more obvious areas. I’ve written about them previously, and while there’s variance between the ministries, the general process is:

  1. You pay a monthly “shared amount” aka a premium. Importantly, this does not go to the health sharing ministry, but to a wallet or escrow.
  2. You end up going to a doctor for something.
  3. The doctor sends you the bill - you upload that bill to the health sharing ministry. Some ministries will negotiate the bill down, others will have pre-negotiated rates with a network.
  4. You have an annual “unshared amount” aka. a deductible you need to first hit before the coverage kicks in.
  5. The health sharing ministry then coordinates the money from people’s wallets/escrows for the remainder of your bill.


This actually doesn’t sound super different from a DAO. People can lock capital up into a pool and then vote on the multitude of service providers to outsource network negotiations, claims assessment, etc. to. Maybe providers join the DAO if they’re in-network and the amount of payment for services is determined by a committee within the DAO and adjudicated instantly through a smart contract. It would probably save some of the admin fees a health sharing ministry usually takes for coordinating all of it.

A second type of insurance product I think this would make sense for is fixed indemnity insurance. This is basically when you get paid a lump sum in cash if you get a condition/disease/an event happens to you. This maxes out at a certain amount, after which the insurance won’t cover you. Fixed indemnity is supposed to be a supplemental insurance product to existing health insurance. 


Fixed indemnity insurance seems like it’s theoretically the most binary form of insurance. If the thing happened to you, you get paid out at X amount if it’s proven. You give the insurance company the ability to access all relevant records that will either prove or disprove you’re covered under a claim. If we had immutable health records or trusted “oracles” (e.g. physicians that confirm a diagnosis) then it should be relatively easy for a fixed indemnity DAO to determine whether the claim should be adjudicated (and farm out questionable claims to trusted third-parties for review).

In theory if this system works, the premium paid to the insurer to normally absorb risk and coordinate these functions would instead stay within the insurance DAOs and theoretically result in cheaper insurance with more transparency in decision making. I think this gets particularly interesting when you start thinking about dollars that payers supposedly get in discounts that should flow to patients (e.g. rebates on high cost drugs), but no one is really sure how those decisions are made. Or in fraud, waste, and abuse where lack of small scale audits amount to lots of lost dollars (maybe members of the insurance DAO can get bounties to identify fraud  in payouts?).

The first wave of cryptocurrency insurance products are running into issues with how governance and claims get paid out, let alone the fact we haven’t had a serious bear market where token prices might drop. I think there’s still a ton to be ironed out before you could ever seriously get people on board with this idea. But an insurance DAO could be the first entity that’s incentivized to actually benefit the members of the pool instead of current insurance companies that optimize for their own margin + shifting costs to their members.


Health Insurance DAOs, a visual


Drug Pickers - A Meta-Model

There’s a hedge fund called Numerai that trades using a meta-model built on top of the models of other data scientists. This document is a way more thorough explanation of how it works, but my understanding is:

  1. Numerai raises a normal hedge fund
  2. It puts out datasets about public companies that are cleaned, regularized, and obfuscated so you don’t actually know the companies the data relates to.
  3. Data scientists build models using this data to try and predict stock performance.
  4. Those data scientists stake a cryptocurrency called Numeraire on their model.
  5. Numerai builds a meta-model that sits on top of all of the submitted models with weights based on how much Numeraire is staked on the model.
  6. Numerai pays out the data scientists based on how correct their model is and how much they staked. If your model sucked, your Numerai is burned (git gud).

The general idea seems to be that the collective of data scientists will build a model better than a small team of in-house ones at a quant fund. It also allows people to participate anonymously + earn based entirely on the merits of the model vs. anything about them as individuals.

The Numerai token seems to just be a mechanism to prevent bad actors from sending random predictions + rewarding data scientists so that they’re incentivized to improve the model and resubmit. 

Numerai seems to be working? Idk I don’t evaluate hedge funds but considering they were trying to build a market neutral hedge fund (keep stable returns when markets go up or down), they seem to be doing well compared to some of the other name brand quant market neutral funds. I see a lot of red and a little green so my monkey brain interprets this as them doing well, but more sophisticated people can tell me if it’s true or not.



It strikes me that you can potentially apply a similar model for drug development? Get the data from animal models + clinical trials, anonymize the asset itself, and then have a community of data scientists build their own staked models to guess which drugs will perform better in future trials. Maybe the data scientists get a % of royalties on the portfolio of assets they’re evaluating, depending on how correct they were/how much they staked.

I think this might solve a few problems. Data scientists seem to be in short supply and hard to recruit to the biotech/pharma world. This article suggests a range of issues from big tech companies offering more lucrative packages + the data scientists doing more “data janitorial” work in life sciences. By doing a meta-model, fewer rockstar data scientists could potentially be working across different pharma companies/pipelines. Plus, you could provide incentives in the form of earning currencies to be staked for cleaning data. And you could have non-life science data scientists potentially working on these problems on the side who might bring different perspectives to the table (this is how RenTech seems to have been successful as a hedge fund). 

There are definitely hang ups here. For one I’m not even sure you could anonymize the datasets completely around the assets. The main issue though is that the stock market provides constant feedback on performance, whereas the feedback on drug development can take years so it’ll be much harder to fine tune models. 


Still, I think the idea of giving structured datasets + having data scientists build models + building out staked meta-models on top could be applied somewhere in healthcare.


Those are some ideas that come to mind, but timeline and feasibility for any of these are still big question marks. If you have thoughts or ideas on these or other blockchain-based projects, let me know. And please don’t shill me your pump and dump coins (...unless I’m the very first person you hit up).

Thinkboi out,

Nikhil aka. “I be on that crypto-night”

{{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.


Let's Keep In Touch

Thank you! Your submission has been received!
Oops! Something went wrong while submitting the form.
close
search icon
close