An ambitious new blockchain & federated learning project

Blockchain in healthcare updates

Owkin, NVIDIA, and King’s College London collaborate on patient data federated learning project

Owkin is creating federated learning technology and is a key member of MELLODDY. They use a blockchain to orchestrate and keep an audit log of the federated learning process, giving participants an even higher degree of confidence their data isn’t used in ways they don’t approve of. They teamed up with NVIDIA (also a member of MELLODDY), who released their federated learning for healthcare framework called Clara this week, an AI developer kit for image and video processing, an edge compute platform (which is synergistic with federated learning).

Alongside Owkin and NVIDIA is King’s College London’s AI Centre for Value Based Healthcare (AI4VHB), which will then work with three universities and four hospitals to form the basis of a federated learning network. The scope seem is initially rather broad:

KCL will use Owkin’s Federated Learning software and NVIDIA’s EGX Intelligent Edge Computing platform to develop research, clinical and operational improvements across a large number of clinical pathways, with cancer, heart failure, dementia and stroke likely areas of early focus.

Given Owkin’s previous work and NVIDIA’s new products it seems likely the initial focus will have to do with imaging data. This line also interested me:

KCL will make this federated dataset available to its researchers as well as a community of Life Science & Healthcare companies that are consortium members within the Innovate UK funded AI Centre.

If I’m understanding this correctly other members of AI4VBH will have the ability to train algorithms on the federated datasets created. In theory this is equivalent to pooling data from a bunch of institutions, but in this case data never has to be shared. As a result this architecture should lower the barriers to training algorithms, and thus gaining valuable clinical insights, on sensitive data. More researchers will be able to gain the benefits of ever larger health datasets (albeit federated!), but without the usual tradeoff of patients sacrificing their privacy.

Today this is confined to a relatively small group of people affiliated with AI4VBH but I can imagine how this architecture could scale much easier than traditional patient data sharing can scale. Lastly, I’m interested in how they govern the algorithms and insights that are generated from this collaboration. For example: if one of the AI4VBH partners finds some novel insights with their algorithms, who has a claim to that algorithm? Who gets to chose when/how/if it’s monetized?

SimplyVital Health launches a blockchain-powered audit API for healthcare organizations

SimplyVital has been building blockchain for healthcare data infrastructure for ~2 years now and it seems they are focusing on enabling value-based care with their technology. To that end they’ve launched an API that makes it easier to record and audit events on a public blockchain, which is valuable if you need an audit log for sensitive data or processes. Also of note: they recently received a grant from the NSF.

Rymedi teams up with Kadena to launch a CBD oil tracking solution

Rymedi builds blockchain and healthcare solutions and is one of the teams conducting an FDA DSCSA pilot. They are also notable for their work in Mongolia. They teamed up with Kadena, which has recently created a new smart contract platform, to launch a tracking platform for CBD oil. The idea being that consumers could scan a barcode and see the journey of a product across a supply chain into their hands, and have greater confidence that they are taking something which is high-quality. The article also makes it clear this new partnership has ambitions to build out more healthcare infrastructure together.

A quick note

Last week I couldn’t find any details on “FantomOperations,” which is working with the Ministry of Public Health in Afghanistan, and Charles Fournier was kind enough to point out to me that was the Fantom Foundation. Much appreciated!

Blockchain, Wikis, and the Ideal Science Machine: With an Example From Genomics

Blockchain and genomics company raised ~£1.05m on a crowdfunding platform

That brings total blockchain and healthcare (non-ICO) funding for the year to ~$38m.

What I’m reading this weekend

Apple is working on federated learning as well

I noticed they were sponsoring a talk at the industry conference NeurIPS. There’s no way that Apple isn’t thinking of how to apply federated learning to healthcare.

Digital Currency Wars: A National Security Crisis Simulation

Harvard Kennedy School’s Institute of Politics held a live simulation of a White House National Security Council meeting and featured participation from an impressive roster of academics and former heads of various federal departments. The simulation covered a crisis set two years out from now where North Korea has dramatically advanced its nuclear program and is actively acquiring raw materials using China's Central Bank Digital Currency as a payment network. In this case the United States cannot exert influence and enforce sanctions like it can today because it has no control over these payment rails.

This kind of future is rapidly becoming a reality! China has been very clear about their digital currency ambitions and it is no secret that North Korea has been hacking crypto exchanges.

It is worth repeating why I highlight this sort of content for my almost entirely healthcare focused audience: geopolitical digital currency conflicts are driving national cryptocurrency and blockchain technology strategies. These strategies help coordinate activity and investment as well as create favorable regulatory environments. This will have huge downstream consequences for the blockchain industry at large.

Kelly Loeffler is the CEO of Bitcoin Exchange Bakkt and soon to be the one of the Senators for Georgia.

Matter Labs announces their very ambitious zkRollup solution

Rollups have made an appearance several times in this newsletter, check out this link for a more in-depth explanation. Matter Labs got a grant from the Ethereum Foundation to build a roll up solution using zero-knowledge proofs and they shared details of their work earlier this week. If implementation goes smoothly this is a big deal. It promises ~2,000 transactions per second with the same level of decentralization Ethereum has today with privacy preserving smart contracts and a cool new programming language. Definitely a project to watch.

I thought this was cool: a new fintech app is giving equity to early users who download the app and invite their friends.

Who will be the first consumer facing digital health company to do the same?

The ergodicity problem in economics

a16z: Programming Biology: Read (r), Write (w), Execute (x)

If you found this newsletter valuable then you can click the button below to sign up for free.

If you’re an existing reader I would deeply appreciate it if you share this with people who would find it a valuable resource. Feel free to reply to this email with comments, questions, or feedback. I also host a blockchain and healthcare Telegram group that you can join.

Happy Thanksgiving from Robert Miller's blockchain & healthcare newsletter

Hey all, happy Thanksgiving if you’re celebrating. I’m thankful for my readers and the many incredible entrepreneurs and innovators driving forward positive change in healthcare using blockchain technology. This week is relatively short as many folks have been away.

Blockchain in healthcare updates

Afghanistan Ministry of Public Health signs MoU to apply blockchain in the health sector

The Ministry signed this MoU with FantomOperations, a firm which I cannot seem to find any details on. Nonetheless, the press released highlighted identifying fake medications and medical records as use cases to be tackled.

An brief interview with Optum’s Mike Jacobs on the Synaptic Health Alliance on blockchain in healthcare and the Synaptic Health Alliance

In this interview Mike Jacobs talks about the Synaptic Health Alliance and adoption of blockchain. After successful first results with provider data management I am keenly watching to see what the Synaptic Health Alliance’s next move is.

KPMG’s China, Japan, and Australia teams launches a blockchain track and trace product

It seems this product has a broad scope and isn’t limited to healthcare, but KPMG’s leaders reference work on going in China with pharma device and medicine manufacturers. I doubt that it is a coincidence that KPMG moved to announce this work a few weeks after President Xi’s announcement a few weeks back.

UCSF researchers using blockchain to share clinical trials lab results

Dr. Laura Esserman, Director of the UCSF Carol Franc Buck Breast Care Center, and her team have been working on a proof-of-concept for how a blockchain can be used to standardize and share clinical trial lab results. It leverages the fairly new Salesforce blockchain and the idea appears to be automating processes that still rely on people to drive down costs. I looked for more details online but couldn’t find any, and I am a bit skeptical given the details shared in the above article. It is generally not a good idea to put any health data on chain because that data will be shared broadly and it is, by nature of using a blockchain, very difficult to revoke.

Previously researchers at UCSF had created a proof-of-concept using a blockchain to prove the integrity of data collected in a clinical trial.

What I’m reading this weekend

Vitalik reviews the progress made in the “Hard Problems in Crypto”

In 2014 Vitalik created a list of hard problems in cryptocurrency space. 5 years later he reviews the progress that has been made. Vitalik highlights three categories of problems:

  1. Cryptographic, highlighting scaling, time stamping, proof of computation, code obfuscation, and problems in hash based cryptography

  2. Consensus theory, largely improvements to proof of work and proof of stake, and 

  3. Economic, discussing how to properly design systems of incentives to achieve desirable outcome, like stablecoins, public good funding, reputation, “proof of excellence,” sybil resistance, decentralized contribution metrics, and decentralized success metrics.

It’s useful to zoom out for a second and think about these problems. A lot has changed in the past 5 years, and indeed a lot of that change has come in the last two, but the hard problems of today remain largely the same. That being said, these hard problems are very much focused on public blockchains and their consumer facing activities. From an enterprise perspective the hard problems holding back adoption have less to do with technical innovation, and more to do with finding clever ways of creating value as well as governing shared infrastructure.

A researcher affiliated with the Ethereum Foundation was arrested by the DoJ for allegedly assisting North Korea in evading sanctions

It should be said that this was a personal venture, not an official Ethereum Foundation activity.

Facebook’s Libra Is Half A Century Late And A Navy Short

What to Consider Before Trading Your Health Data for Cash

Why Google’s Move into Patient Information Is a Big Deal

Breaking Mimblewimble’s privacy model

If you found this newsletter valuable then you can click the button below to sign up for free.

If you’re an existing reader I would deeply appreciate it if you share this with people who would find it a valuable resource. Feel free to reply to this email with comments, questions, or feedback. I also host a blockchain and healthcare Telegram group that you can join.

Veterinary industry blockchain credentialing solution & support for blockchain grows globally

Blockchain in healthcare updates

IBM and Veterinary Industry Group form blockchain credentialing group

IBM is collaborating with VetBloom, the digital learning ecosystem from Ethos Veterinary Health, to create a “group” for using a blockchain for credentialing in the veterinary industry. Critically this credentialing solution is focused on “micro-credentials,” like learning a particular new skill, instead of “macro-credentials,” like a degree.

In short each time someone is issued a micro-credential then a verifiable and permanent record will be created on a blockchain that corresponds to that credential. The idea is that individuals will be able to point to these records on a blockchain to prove their qualification to employers, making it easier to apply for jobs for the applicant, and reducing the costs to verify credentials for the employer.

An interesting contrast to note here is with other credentialing solutions, which have enabled the reuse of one parties’ verification of a set of credentials. The workflow, at a 10,000 ft level, is something like this:

Institution issues credential —> intermediary requests relevant documents about this credential to verify it —> record(s) of this verification process are posted to a blockchain —> others can leverage these record(s) instead of repeating the same verification process themselves

The solution proposed by this veterinarian group goes directly to the source and has the relevant institutions issue credentials directly on a blockchain:

Institution issues credential —> a record of this is posted to a blockchain —> others can leverage this record to verify credentials

Both are more efficient than the status quo. But all other things being equal if you are given the choice of two solutions to verify a credential, where one is directly from the source, and the other is through an intermediary, you’ll default to going directly to the source. One has to wonder which (macro) credentialing authority in healthcare will start to issue credentials on a blockchain first and which solution they’ll partner with.

Chinese pharma company announces blockchain solution for supply chain finance

Details here are sparse, but it seems this is a blockchain solution for making payments in a pharma supply chain more efficient. They partnered with Zheshang Bank, which had previously securitized loans using accounts receivables, perhaps signalling a future direction.

IBM exec on what’s holding back blockchain solutions in pharma:

“It is very hard to get companies that compete with each other to work together,”​ he said, adding: “And then, if we get them to work together on a single problem, and ultimately we are proven wrong on the efficiency or effectiveness of the solution, that’s a lot of work for nothing.”

This echos the sentiment expressed by industry leaders as well that governance problems are more difficult than technical problems.

An interview with Molecule’s Tyler Golato and their research partner Rotem Petranker on their soon-to-be released product Molecule Catalyst

What I’m reading this weekend

A good readout of effects of President Xi’s speech on blockchain by a savvy crypto VC

The article speaks from a crypto-first, blockchain-second perspective and they highlight four larger consequence we can expect to play out in China after Xi’s speech endorsing blockchain:

  1. Crypto-native companies in China will face less regulatory pressure

  2. There is more room for entrepeneurs to experiment in legal “gray zones”

  3. Blockchain initiatives, policies, and funding will come top-down from the government and corporate boardrooms.

  4. Enterprise blockchain providers emerging and selling “blockchain as a service” to businesses and governments.

More global blockchain support:

South Korea’s ministry of science and technology plans to invest $380m for blockchain R&D from 2021 to 2026

The President Elect of the European Commission highlighted blockchain briefly in her agenda for Europe

The European Investment Fund announced this week the launch of its €100m blockchain and AI program

I hope to see the same top-level support for blockchain initiatives in the US.

Federal Reserve Chairman Jerome Powell tells Congress that they are evaluating whether a digital currency makes sense

Specifically, they looked at it previously and came to the conclusion that the benefits of such a project were less appealing for the US, where cash is prominent, the “payments landscape is highly innovative and competitive,” and a digital currency would pose many important policy questions. It’s a good read on the various issues and detailing how the Fed is thinking about this at a high level.

While reviewing this, and the above stories of other global activities, I was struck by how much the overton window had been moved, particularly by the announcement of Libra. These proposals and headlines above would have been unthinkable not too long ago.

Bison Trails, a Libra Association member, raised $25.5m Series A

EY publishes a paper comparing the total cost of running a private blockchain to posting private transactions on a public blockchain

Using a public blockchain you incur transaction fees that are necessary to get your transactions included in the next block. These aren’t that costly (especially if you’re willing to wait awhile) but it is rather expensive if you want to post private transactions on a public blockchain today because doing so uses advanced cryptography that generally requires a lot of extra computation.

In contrast running a private blockchain means you’ll incur the one-time costs of setting it up and on-boarding members as well as the relatively low on-going costs of running nodes and validating transactions. EY’s paper looked at the tradeoffs and economics involved to determine when each solution was more prudent. Some takeaways:

  • Posting private transactions to a blockchain today is very expensive but the next iteration of EY’s zero-knowledge proofs, supposedly launching soon, will drive this cost down by ~90%.

  • Even with the more expensive private transactions today if your blockchain solution has relatively low transaction volume it is less expensive to use private transactions on a public blockchain due to the high one-time costs associated with setting up a private network.

  • There’s a point that EY identifies close to ~2,000 transactions a day where it becomes more efficient to use a private network over their future state (cheaper) ZKPs.

The model EY creates make a series of assumptions around costs that may or may not be true in any given situation. But the general heuristic of “public blockchains are more efficient for low transaction volume and private are better for high transaction volume” seems reasonable.

Visa’s R&D Arm develops a TEE & blockchain based system for parsing sensitive financial data

In a paper published by Visa’s research and development arm, researchers describe a system called LucidiTEE. It outlines a system for sharing sensitive personal data on a blockchain, crunching that data within a trusted execution environment (TEE) and using history-based policies to ensure that each of the parties receive an output of the computation. (The system’s name is a combination of TEE and the word lucidity).

Easy to see how this could be applied to healthcare as well.

Podcast: Vitalik Buterin on The Portal

If you found this newsletter valuable then you can click the button below to sign up for free.

If you’re an existing reader I would deeply appreciate it if you share this with people who would find it a valuable resource. Feel free to reply to this email with comments, questions, or feedback. I also host a blockchain and healthcare Telegram group that you can join.

Novartis talks blockchain on CNN & Google's aggregation of health data comes under scrutiny

Blockchain in healthcare updates

Novartis execs live on CNN Money Switzerland talking about blockchain

Some highlights:

  • Believes that near-time value in use cases closer to supply chain, and but there is long-term value in patient data use cases

  • Stressing their desire to go live with production use cases in 2020

  • Highlighting the IMI blockchain and healthcare consortium, which has 11 pharma partners and 28 companies participating in total. It is set to launch in the coming months and I’m keen to see what they tackle

  • When asked about how they compare to their peers they didn’t directly answer the question but talked about their work on digital identity and verifiable credentials

  • Emphasized the need for an ecosystem in blockchain solutions

Cool to see blockchain and healthcare getting the spotlight like this!

NYT highlighting Oasis Labs and their work with Nebula / Kara

Oasis is building tools to let people control their data by leveraging blockchains and special hardware called trusted execution environments (“TEEs”). In short, TEEs provide an isolated area to store sensitive data and run software on it, and this area is both private and resistant to tampering even if you have access to the physical hardware. Your mobile phone likely uses a TEE to store your biometrics or financial data.

Oasis hopes to marry the confidentiality and integrity of TEEs with smart contracts to give people real “control” over their data. The article highlights a pilot being done with Nebula Genomics, a blockchain and genomics company, where users will retain control over their genomic data while still enabling Nebula to run specific analysis on their data without revealing it to Nebula. Another collaboration is Kara, where patients have the option to share retina scans and other medical data with researchers to train machine-learning models to recognize disease, and they hope to test what kind of incentives patients respond to.

There are several projects taking the approach of using TEEs to give users more autonomy and privacy. Enigma, is another project and just last weekend I was working on a use case fairly similar to Kara using Engima’s tech instead of Oasis’. Hyperledger Avalon, is a “framework” and does not specify the implementation of particular technologies like its peers, but has similar goals, though Avalon’s scope goes beyond TEEs.

Overall my personal assessment is that these technologies are nascent, and have work to be done to make them production grade, but have great potential.

Korean EHR startup signs another MOU

MediBloc is one of the 2017 healthcare ICOs that is still around and occasionally putting out headlines. They tout this as their 12th MOU, which is nice, but it occurred to me that I’ve never heard anything more about what has happened with the past 11 MOUs that they have signed. Take that as you will.

Podcast: yours truly talking about blockchain and healthcare, federated learning, and bioethics

Podcast:, an open-source platform to run “private AI advisors,” touching on blockchain and federated learning

What I’m reading this weekend

Google’s secret project Nightingale, carried out with Ascension, gathered the personal health records of millions of American

The initiative moves tens of millions of personal health records to Google’s cloud, where they will have the ability, indeed the explicit aim, of analyzing huge amounts of identified personal health information. The WSJ, citing anonymous employees, reported that as many as 150 Google employees have access to this patient data, and reported that patients and physicians were not informed of this work. A day after the WSJ’s report broke they also reported that the HHS has opened a probe into this arrangement, though it’s not immediate clear to me when that probe was started.

Google defended this work as in line with industry standards, HIPAA compliant, and subject to a Business Associates Agreement they have with Ascension. But given Google’s history, which a charitable reading would characterize as misleading the public on their practices with our data, it is hard not suspect there is more going on here than Google lets on. Moreover, without seeing the contents of their BAA we won’t know the limits of Google’s work. For example BAAs often times include a provision allowing the business associate (Google) to deidentify data, which in turn means that data would no longer be governed by HIPAA and provided HIPAA’s protections.

Google has built their brand on an image of relentless curiosity and built their business on a relentless march to gather more and more data about us in the face of growing privacy concerns. This history, along with other notable tech mishaps, has led to a huge trust deficit where only 10% of consumers are willing to share their health data with tech companies. Against this backdrop it is no surprise that people are turning to technologies like blockchain and trusted execution environments to fill that gap.

Are Patients Electronically Accessing Their Medical Records? Evidence From National Hospital Data

A new study in Health Affairs found the following, bolding mine:

On average, hospitals gave 95 percent of discharged patients access to view, download, and transmit their information, but only about 10 percent of those with access used it—levels that were stagnant during the study period.

Pew research on American’s and privacy

More telling statistics

81% of the public say that the potential risks they face because of data collection by companies outweigh the benefits

81% say they have very little/no control over data companies collect

79% say they are very/somewhat concerned about how companies use(s) the data collected

59% say they very little/no understanding about what companies do/does with the data collected

These questions were also run with the government being the focus, not just companies.

CoinShares releases an epic 138 page report on 2019 Crypto Trends

There’s a lot here, and I’m not through the whole report yet, but it’s a great read.

Multi-collateral DAI is coming tomorrow

How top health websites are sharing sensitive data with advertisers

A good blog from the Coinbase on how they think about proof-of-work

A Crossroads, not an Island

After this podcast with Balaji Srinivasan I was interested by Glen Weyl’s ideas on identity and how current market mechanisms fail to account for the robustness of social relations, so I’ve been digging into that idea this weekend. I read this paper when it came out and didn’t have much of a reaction to it, but it became much more interesting after hearing Glen speak to the concepts using examples on the aforementioned podcast. I’ll try to write something about how this is relevant to healthcare after I digest it a bit more.

If you found this newsletter valuable then you can click the button below to sign up for free.

If you’re an existing reader I would deeply appreciate it if you share this with people who would find it a valuable resource. Feel free to reply to this email with comments, questions, or feedback. I also host a blockchain and healthcare Telegram group that you can join.

Compliance by design and ETHWaterloo

Blockchain in healthcare updates

Blockchain Compliance by Design: Regulatory Considerations for Blockchain in Clinical Research

A dense and important article detailing a list of regulatory considerations for blockchain’s usage in clinical research. The authors did a good job highlighting the various different areas developers and policy makers need to pay attention to, and provided a series of recommendations:

  1. Education for researchers and regulators

  2. Engagement between researchers and regulators

  3. Sandbox for design and development

  4. Administrative blockchain pilots

  5. Clinical research pilots

  6. Clear and consistent data privacy protections

  7. Interoperability standards

There is an interesting follow up article to be written about the unique compliance challenges for blockchain consortia. In consortia organizations join together to use shared infrastructure, which has great business benefits, but it also means that each individual organization gives up a degree of control over this infrastructure. It’s not hard to see how this could be problematic if another actor does something non-compliant, which will inevitably happen as these networks scale. Managing these non-compliant actions and reducing liability when using shared infrastructure will be challenging.

Adoption of AI and Blockchain at HHS: Interview with Jose Arrieta, US Department of Health & Human Services (HHS)

Another look at Jose Arrieta’s project HHS Accelerate, a very successful deployment of a blockchain used for procurement. Jose was appointed to CIO of the HHS in May of this year.

I spent this weekend at ETHWaterloo, a blockchain hackathon

My very talented colleague Tyler Farnan and I teamed up to implement machine learning on-chain within Enigma smart contracts (which are also private!). That means you could train a machine learning model on-chain (or plug in an existing model), charge people to use it, pay people to contribute their data, manage it with a DAO, etc. We used 1upHealth to be able to pull health data, but used Synthea to generate thousands of synthetic FHIR standardized health records, for creating an algorithm that could tell if you were at high risk of a particular heart disease (ATTR-CM).

Long story short we ran into some esoteric bugs with Rust that kept us deploying that algorithm on-chain, but with time, and more knowledge of Rust, it is possible to do so. There are limitations today (e.g. your models need to be no larger than ~1MB, you can’t use the most popular ML tools), but it blew my mind that this was even possible. We won an award from the Enigma team for our work too.

Another team independently had the same idea and implemented it as well! The uFlo team pushed data from a neuro infrared spectroscopy hardware (fNIRS) headset into a machine learning algorithm on Enigma to identify when someone was in “flow.” Because this algorithm was in an Enigma smart contract the whole process of submiting your data and getting an output from the model was private and trustless. Super cool. Another healthcare related project was CypherBabies. There’s a lot of other interesting non-healthcare stuff, with a lot of teams building DeFi applications or using NFTs.

Anyway, nothing gets me more excited for the future of these technologies than hackathons. It is amazing to see what a team of people can build in a focused weekend. Part of the reason why people can do so much is because of all of the work that has gone into new tooling and infrastructure. Almost everything built at Waterloo this weekend would have been next to impossible to build a year ago, let alone before that. And in a year the tooling and infrastructure will be even better still, which begs the question of what cool stuff people will be building in the future that isn’t feasible/possible today.

What I’m reading this weekend

Optimistic roll-ups vs zk-rollups

Roll-ups are an approach to scaling blockchains that have received much attention in the past weeks, including in this newsletter. They are a “layer 2” solution, but are different than other layer 2 challenges by storing all relevant transaction data on-chain and moving only execution of those transactions off-chain. That seems a bit counter-intuitive (at least it was for me), but it works, and shows promise in scaling the Ethereum main net in the short term. This article compares Optimistic roll-ups with another form of rollups using zero-knowledge proofs, zk-rollups.

The Story of Modern Cryptography

Shafi Goldwasser is an important researcher who boasts won multiple awards at the highest level of cryptography, and was the lead author on the paper that proposed zero-knowledge proofs back in 1985. In a talk at a VC fund’s summit she recounted the story of modern cryptography.

Balaji Srinivasan and Glen Weyl on Identity, Governance, and Radical Markets

Balaji and Glen are two extremely sharp thinkers that I follow. Their conversation on this podcast was wide ranging on blockchains, crypto, identity, governance, and more; it is very dense and the kind of podcast that rewards multiple listens.

Hackers can use lasers to speak to your phones or home devices

If you found this newsletter valuable then you can click the button below to sign up for free.

If you’re an existing reader I would deeply appreciate it if you share this with people who would find it a valuable resource. Feel free to reply to this email with comments, questions, or feedback. I also host a blockchain and healthcare Telegram group that you can join.

Loading more posts…