Andrei Pop is the founder and CEO of Human API, a universal health data platform that enables developers to easily build the next generation of healthcare applications. He is also a serial entrepreneur, speaker, and expert in the field of healthcare data.
Andrei's focus on health data began while developing real-time movement feedback systems through wearable sensor technology. He has presented his work at numerous conferences, including the Cleveland Clinic, International Interface Summit, Le Web, and Singularity University. I talked to him because his company is in an interesting space and is a great illustration of REST APIs, a topic closely associated with the JASON recommendations I discussed in my last post.
Mark Braunstein: Andrei, what was the motivation for starting Human API?
Andrei Pop: Our health data is stored in a lot of disparate datasets. Due to very difficult technical challenges, we rarely get the opportunity to pull it all together and create a clear, quantitative picture of an individual's health. Over the years many have proposed that an answer to freeing health data might be to let patients control it and make sharing decisions directly. That's the basic idea behind Human API.
[A more open approach to health IT: A Serious Proposal For Healthcare IT Interoperability.]
Braunstein: Suppose I'm a patient with a device at home with which I can measure my blood pressure, serum glucose, or whatever. Tell me what Human API can do with that data.
Pop: We make it simple for your physician, a nurse, a researcher, or perhaps a caretaker to retrieve this data. With your explicit permission, your care providers can retrieve this data through our simple, secure API. We normalize your blood glucose data and many other forms of health data so that your care providers can turn it into actionable, meaningful insights.
Braunstein: Is this just for device data?
Pop: No, we want to be the easiest way to integrate health data from anywhere. That includes more traditional clinical data, labs data, genetic data, and device data.
Braunstein: Once it's aggregated on Human API, what do you do to facilitate sharing of that data?
Pop: We normalize the data into a standard model so that the applications that get the data can more easily utilize it. Without this normalization of schemas, these applications would have to directly query the individual data sources and deal with their own unique and often obscure interfaces and data models. We then display the data through simple RESTful APIs, making it much easier to ingest and then build applications and workflows around it.
Braunstein: Can we see a specific example of this API-based query?
Pop: In Figure 1, you can see the API queries we defined for blood pressure. In an actual query, each would be combined with the access token. Here's an example of that with the token in bold at the end:
The patient identity is confirmed by the identity they create when they use whatever product integrates with the Human API infrastructure. An example might be logging in to a portal before they click to authenticate and share data via Human API. They can then choose which data they wish to share by checking it on a web page.
Braunstein: How is adoption going?
Pop: It has been overwhelming, actually -- we currently have much more demand from application developers than we expected; we are hiring like crazy to keep up with demand. What is cool to see is that some of our customers are startups and others are very large health IT vendors or provider groups. Everyone is seeing the value.
Braunstein: Since we did the interview, the JASON report had recommended an approach quite similar to yours and the FHIR standard has been gaining a lot of traction. Both involve a standard data model and API-based access to data. What is Human API's view of these new developments?
Pop: I was pretty thrilled when I read that paper. I kept thinking, "These guys get it." Given how important it is to solve this problem, it's great to see others tackling it as well.
The owners of electronic health records aren't necessarily the patients. How much control should they have? Get the new Who Owns Patient Data? issue of InformationWeek Healthcare today.Mark Braunstein is a professor in the College of Computing at Georgia Institute of Technology, where he teaches a graduate seminar and the first MOOC devoted to health informatics. He is the author of Contemporary Health Informatics (AHIMA Press, 2014) as well as Health ... View Full Bio