Gemma is Managing Editor at AIMed, with responsibility for engaging and growing the AIMed community and to highlight stories of health AI in action. An experienced science graduate with a background in veterinary and nonprofit sectors, she also volunteers as a Wish Granter for Make a Wish UK.
“There had to be a better way – for patients, caregivers, providers, everyone”
Equideum’s Heather Flannery on building a rich ecosystem where truly transformational and explainable AI can be developed and can thrive
You started your career as a web developer. What led you to become involved in medical AI?
I’ve always been a serial entrepreneur. I started in tech that was fairly general in scope and transitioned into health-specific tech almost 20 years ago. I made the move in part because of my personal experiences as a caregiver and a patient trying to navigate health systems. The inadequacies and system difficulties I encountered drove me toward wanting to find better solutions. There had to be a better way – for patients, caregivers, providers, everyone.
What motivates you to make an impact on health equality and global population health?
When thinking about health equality or, more specifically, health equity and global population health, there is an overwhelming need, not only in the United States but also across developed and low and middle income countries, to improve health for everyone – not only those that can afford the best care. Currently, most health advancements and research are focused on a very narrow subset of the population. As one example, AI tools being developed are not generalizable to all populations. We aren’t exactly sure who will benefit from the AI trained models and who may be harmed with poor care recommendations. By focusing on health equity and data liquidity for all populations, which is what we do at Equideum Health, we can bring global population health to a new level with better, more explainable AI. That will let us know where it is and isn’t generalizable in the wider population. We’ll see where we have more work to do in better inclusion in clinical research and data gathering for better AI and treatment options for all.
What are the main challenges you face?
One of the challenges we face is that the current health tech system was modeled after what was successful in tech generally over the last 30 years – build an app, get it to market, then scale. This is one of the reasons fax machines are still being used in healthcare in the United States.
Product-market fit only gets us incremental progress – “faster horses” as Henry Ford said about the product-market fit when he introduced his Model T. All the key stakeholders, from large corporations to venture capital firms, to the start-up community, are oriented around small incremental changes.
But we know from our current experience that transformational changes require a large, concerted effort among all stakeholders – big companies and small companies, public and private groups, and most of all patients and providers. It takes a lot of capital and time to introduce these new ideas and develop new systems, not just new products for the legacy system. You can’t simply build tree after tree and expect to get a forest ecosystem. There needs to be widespread change introduced across several areas in a short timeframe. We regard people’s health, and the data that underlies it, as a new data economy – one which must be inclusive of the patient. This takes a patient/consumer focused effort at the same time as an enterprise focused effort – what we call a data integrity & learning network or DILN. These two sides need an interface or marketplace, which we are creating in our Equideum Exchange. This is the rich ecosystem where truly transformational and explainable AI can be developed and can thrive.
To what extent do you see artificial intelligence helping to solve these challenges?
Artificial intelligence enables more rapid processing and knowledge discovery of everything from administrative processes to advancing health diagnoses and treatments to finding better ways to deliver those treatments to a wider array of individuals more precisely and on an individualized basis.
Not everyone is the same. The basis of a person’s health status for treatments and wellness should come from not only what is in the health record but also from health relevant data such as environmental data and social determinants of health. Ideally, all this data can be incorporated into a person’s individualized diagnosis and treatment plan. This requires managing a lot of information, and AI is a great tool for allowing this to happen at scale with the necessary speed.
What advice would you give someone starting their career in medicine or medical AI?
Those starting a career in medicine probably want to pay attention to AI and other technologies that are becoming available because those will be tools that they will be utilizing for diagnosing and treating their future patients.
For those pursuing medical AI, they should be careful not to be too narrow in their vision. Currently, less than 1% of medical AI has been clinically validated. Those numbers need to be brought up, not just by developing new algorithms, but by being able to train those algorithms on the right kind of data, being able to explain where/how that data was trained, being able to implement it in a privacy preserving way, and being able to clinically validate that the AI results improve outcomes.
We need to think about the whole system and the individuals involved. At Equideum Health, we approach that by using blockchain for the underlying “explainability” layer for what data the AI was trained upon. We also rely on privacy preserving technologies – everything from hardware solutions like trusted execution environments to software approaches like zero knowledge proofs and homomorphic encryption. In a healthcare setting, this makes it possible to have data about a person available for clinical treatment or for research without ever exposing that data. So, those going into AI need to consider more than one-off programs that may be relatively easy to put together but that never actually get put into practice to improve health outcomes.
If you could return to the past, what would you change or do differently?
I spent a number of years in the healthcare and health technology areas with a robust but very specific focus on the areas of obesity and related issues. With the emergence of new technologies that can do so much more, I’ve realized that it has become critical to look across the commonalities of all health issues and not work in one narrow area. There tends to be pressure to focus on one use case, but if you can’t create something that has applicability across use cases, it ultimately will be another niche tool that has no generalizability. So, thinking broadly across use cases and not focusing on an isolated use case is something in retrospect that I would go back and consider in my approach to driving solutions more effectively and efficiently.
To learn more about creating tools that drive solutions, attend AIMed’s Global Summit 2023 … book now! https://ai-med.io/all-events/global-summits/aimed-23/
Heather Flannery is the Founder and Chief Executive Officer of ConsenSys Health. She chairs the IEEE SA Open P2418.6 Standards Development Working Group (blockchain in healthcare and life sciences), served as FY19 and ’20 Co-Chair and FY21 Chair of the global HIMSS Blockchain in Healthcare Task Force, Chairs the Healthcare Interest Group at the Enterprise Ethereum Alliance (EEA), and is an Associate Editor of the peer-reviewed journal, Frontiers Blockchain for Science. She is also Co-Founder and Board Chair of Blockchain in Healthcare Global (“BiHG”), a 501(c)6 trade association organized under the IEEE ISTO. Ms. Flannery has served as Industry Faculty for the United States Department of Health and Human Services Office of the National Coordinator for Health IT (US HHS ONC). She is an active consultant, advisor, and keynote speaker. Equideum (formerly ConsenSys Health), a ConsenSys Mesh portfolio company and ConsenSys partner, builds Web3 person-centered healthcare and research networks called Data Integrity and Learning Networks (DILNs).