Many companies tout the promise of population health, or the ability to influence the health outcomes of a group of people. But the process of achieving this goal, from collecting data to changing personal behavior, can be challenging.
In a panel discussion at MedCity INVEST Pop Health, the leaders of the two companies discussed how to integrate population health data and implement effective solutions.
Early in his career, Dr. Anil Jain, Chief Innovation Officer of Innovaccer, opened his own primary care clinic. At first, he mostly considered one for each patient at a time.
“Until we started collecting this data from electronic medical records, did we really start to think about what the total population is like?” he said. “Can we see all 236 patients instead of one of my diabetic patients?”
Of course, treating patients as a group requires weighing the pros and cons. For example, how do you aggregate the health information of multiple people while still protecting their privacy? At the same time, how do you keep track of who is who, because there is no patient identifier in the United States?
Jain said there are some identity management solutions, such as asking patients to verify their identity, or building a system that can help match certain details, but “there are still a lot of problems with identity management and single patient records.”
In addition to the identity of the patient, there are other things that need to be checked before matching the data. For example, if a doctor writes “CP” in a note, do they mean chest pain or cerebral palsy?
“Many innovators spend a lot of time cleaning data and using the same terminology as another set of data so that they can begin to understand it,” he said.
Another difficulty is actually to make this information operable. Some companies have turned to machine learning as a method, but Jain and Patientory CEO Chrissa McFarlane both expressed caution. Although hospitals and payers are using many of these tools, “the emerging AI challenge lies in fairness and prejudice surrounding AI,” McFarlane said.
She added that in teams that develop software tools to improve results and understand when to use the data, it is important to have more diversity of ideas.
As far as Jain is concerned, he agrees that caution is necessary, and that artificial intelligence should be used primarily for simpler tasks, rather than complex decisions that doctors will make.
In the final analysis, “this has nothing to do with data, or even analysis, but about changing people’s behavior,” he said.
Photo credit: elenabs, Getty Images
