“You have to have a vision for data. It’s all about the data.” Andrade said.

So said Amy Andrade, former assistant vice president of research at Meharry Medical College and founding director of Meharry Medical College’s Data Science Center, during a recent podcast on Healthcare Strategies.

The specific topic was the social determinants of health and the extent to which, in Andrade’s view, the critical data for healthcare providers to have when it comes to patient health is the data found outside the doctor’s office.

“Our health isn’t taking place in the doctor’s office – it’s where we live, where we work, and where we play,” Andrade explained. “You have to step back and look at those influences. If providers and health systems don’t have access to this information, then they’re not seeing the true picture of what’s influencing health. You can’t just look at it in a siloed fashion.”

The goal of having a holistic view of a patient’s health, of course, involves gathering a lot of data, so Andrade argues that “(h)ealth systems, small or large, need to have a data plan in place. They have the clinical data, and now they need to think about how they’re going to use the socioeconomic data. If you’ve got 75 percent of the information, you’ve got to just jump in. You can’t wait perfect it, because then you’ll miss that opportunity.”

“Jumping in” is where AI comes in, Andrade says, because as an organization gathers data it needs to be thinking simultaneously of how best to organize and use it.

“You’ve got to look at your particular health system and what’s plaguing you. For instance, in urban areas it’s the prevalence of underlying chronic diseases. In other places, it may be cancer,” Andrade said.

Next, she said, “you’ve got to ask yourself why. If you peel back the onion, you’ll get to the root cause. So now you have these large amounts of data, and that’s where AI comes in.”

Ultimately, building an AI model that meets the needs of your organization will likely be a process of trial and error, Andrade said, but the goal of developing a holistic understanding of a patient’s health circumstances will be key for every entity looking to improve patient and population health.

“It takes some time, because each model performs differently depending on the data that you have. It is the quantity of data that comes in and the artificial intelligence intersecting with the information you have available,” she said.