We shouldn’t underestimate the power of patient data.
Every data point from clinical studies and patient medical records provides fuel for researchers. The more fuel they have, the greater the chance for life-saving breakthroughs.
As we sit here today, the research ecosystem isn’t working to its full potential. Valuable datasets are siloed and disparate, making it hard for researchers to access them. Patients aren’t empowered to take ownership of their own data; the power of data is in the hands of a few large entities. Companies are profiting from sharing patient data with third parties, creating concerns over patient privacy and data ethics.
How can we unify the system around the ultimate goal of better health outcomes? How do we transition from a data re-selling economy to a patient-benefitting insights economy?
It will only take a few ingredients to discover a better future: a collective mindset shift, some creative technology and nonprofits embracing a role as data stewards. We’re not that far from a future where health data is managed by mission-based co-ops or trusts. Patient data will be a sustainable resource that continually generates patient-benefitting research insights.
Patient advocacy and non-profit health organizations will be the “credit unions” — responsible fiduciaries who connect patients, researchers, hospitals and other institutions. When patients give permission, these organizations will act as stewards that manage the secure, responsible, use of health data on behalf of their patient members.
And best of all — with the help of artificial intelligence and machine learning — all of this can be done in a way that puts patient privacy first and ensures patients still have ultimate control. It’s a winning formula for everyone involved, and patient advocacy and non-profit health organizations have the power to make it a reality.
I’ve had the pleasure of working closely with MIT Professor Sandy Pentland, a world-renowned data scientist. We collaborated on a book, Building the New Economy: Data as Capital, which paints a picture of user-centric data ownership, where data trusts can empower people and communities. Data control must be democratized, instead of allowing a few large entities to control the vast majority of this valuable resource.
The Mozilla Foundation has also become a champion for data trusts, data cooperatives and other efforts to shift the power dynamics around data. They’ve outlined guidelines for data governance and stewardship as we build this new future.
But why hasn’t this thinking taken hold in the medical community? It’s a little complicated:
A patient data economy already exists, but it’s not the mutually beneficial version I just laid out.
There are a handful of companies that offer “free” patient data registries for patient advocacy groups and non-profits. Free rarely means free; these companies essentially act as data brokers and profiteers.
Patients agree to allow researchers access to their data, which seems great on a surface level. But in order to turn a profit, these companies share this data with outside parties for secondary and tertiary uses.
This data can end up anywhere, given how easy it is to forward and move datasets to anyone in the blink of an eye. It can even be traded on the dark web. Even if health data is de-identified, there are still significant concerns over patient privacy. The door is wide open for re-identification and misuse.
Consider a traditional asset, such as land or any natural resource. These resources are finite; the more I have, the less there is for you.
Data is also an “asset,” but of a different kind. When a patient shares their medical record, it can be used by one research institution and then another, for different purposes, without losing its value. We all benefit from greater access to patient data.
HIPAA, the Cures Act and other policies have given patients more authorization over how their data is shared. Still, the general sentiment is that large organizations are the ones with true “ownership” of the data.
Healthcare systems, major PIs and labs control a vast amount of valuable health datasets. Although every organization says its mission is to move research forward, it sometimes feels like they prioritize protecting their “assets,” aka the data sets, over the collective approach of curing disease.
Creating a patient-centric data economy means flipping this standard on its head. Pentland summarized it well when he said we must move, “from an individualized, asset-based understanding of data control, to a collective system based on rights and accountability.”
Patient advocacy and non-profit health organizations are the best fiduciaries for a patient-centric data economy.
We need an alternative to this institution-based model that’s littered with data brokers and profiteers. Patient advocacy organizations are the perfect candidates to drive a new, secure and privacy-preserving patient data economy.
Why patient advocacy organizations? It’s because their mission and vision are targeted toward ending a specific disease type. They’re laser-focused on that goal. These groups are coalescing disparate researchers needs to prioritize the data that supports mission-based research. The patients’ voice is baked into everything they do.
Hospitals have too much on their plate; their mission is to treat thousands of different conditions and help patients get better. We can’t trust companies that tie their profitability to data proliferation and brokerage.
So that leaves patient advocacy and non-profit health organizations. They’re the connectors of the ecosystem and already have the infrastructure in place for the management and sharing of patient data.
Just look at what some next-generation patient advocacy organizations are already doing,
The Fatty Liver Foundation (FLF) is coordinating the development of an AI Fatty Liver Risk Stratification Tool that uses machine-learning technology to help patients assess their risk for nonalcoholic fatty liver disease (NAFLD) and its more advanced form, nonalcoholic steatohepatitis (NASH).
The Kidney Cancer Association (KCA) is using similar AI/ML technology to turn a high volume of rich, valuable kidney cancer data into a comprehensive database that researchers can easily — and compliantly — access. This type of patient-centric data innovation makes these groups the right entities to move research forward.
I couldn’t make a data trust or data cooperative on my own. But if I have 1,000 other Anne Kims willing to pool their data, this vast dataset becomes something valuable that can fuel research. The strength is in the numbers.
How do we actually do something with this data? That’s where patient advocacy and non-profits operate as the managers of the cooperative or trust. Patient advocacy organizations would receive data donations from patients and coordinate the secure use of that data for researchers. Every patient advocacy group would manage this process for their disease type, and it would become really clear from a research perspective where you go for particular types of data.
It would be very similar to how credit unions function, except for data: member-owned, non-profit cooperatives that are required to operate with their owners’ interests at heart.
How would these data stewards securely share patient data? It wouldn’t be like the traditional “lending” that credit unions do with money. Instead, patient-centric groups would upload health data into a secure space or enclave. With the use of artificial intelligence and machine learning models, we can generate insights for researchers without actually “moving” the data out of its secure vault. More on that in a moment.
What ties this whole system together is a mutual benefit. When data is locked away, it doesn’t do any good. If we enable secure, patient-centric data exchanges, we’ll be able to glean exponentially more insights than we ever have before.
How can we find treatments for rare conditions? What about treatments that work for every demographic? How can we create personalized treatment pathways and prevention plans that are effective for each patient’s unique needs?
We could get answers to all of these questions. Whether you’re a disease survivor, a patient’s loved one, or just someone who recognizes the value of their data, you can make a difference.
First, we’ll need to move in a few directions:
The right technology is essential to the data lending process. Patients must be ensured their data will be safeguarded and not duplicated infinitely. All uses of health data must be known and auditable, and overseen by a mission-based organization. That’s why we need to steer towards agnostic software and service providers — instead of data profiteers.
Machine learning and analytics on federated data can help us do computation data analysis without actually moving any data. Clinical researchers must be able to run queries on patient datasets with the confidence that this data can’t be copied and pasted, emailed, forwarded, downloaded, or end up on an unauthorized server. Data must stay in an auditable and cryptographically contained enclave; a researcher enters a query, and the machine learning models come back with insights.
A patient-centric data economy is a little abstract because we haven’t really seen it before. We need to find ways to educate the entire medical community on this concept. People won’t participate in something if they don’t know it’s possible. Patient advocacy and non-profit health entities should be beating the drum for this philosophy.
The patient-centric data economy is a new frontier where disease-ending insights for everyone are at the forefront. That means we’re still establishing a foundation for data rights, governance and stewardship. We have to continue this nuanced conversation because the context could change almost every day. For a more cooperative, distributed system, we need parameters to keep that infrastructure in place.
It’s exciting to think about what a patient-centric data economy could do for the medical community. Patient advocacy and non-profit organizations have a massive role to play, and plenty to gain. The easier and safer we make it to exchange healthcare data, the greater the chance we have of generating disease-ending insights for everyone.
About Anne Kim
Anne Kim is the co-founder and CEO of Array Insights, formerly known as Secure AI Labs (SAIL), which is based on her graduate work at MIT. She worked with world-renowned Professor Alex “Sandy” Pentland on federated learning and blockchain solutions for clinical trial optimization using Open Algorithms (OPAL). Outside of her research, Anne has led a number of different projects in computer science and molecular biology and cyberbiosecurity work with the EFF, ACLU and DEFCON.