Better Health Through Records
A physician walking into a room with a manila folder filled with medical records is as much a part of the image of a doctor as the white lab coat and a stethoscope. In the future, the coat and the stethoscope will likely remain, but that manila envelope will soon be replaced by a computer or tablet, if it hasn’t already happened.
Sriraam Natarajan hopes to make that computer less a repository for electronic health records and more an active member of a physician’s team.
An associate professor in IU’s School of Informatics and Computing, Natarajan hopes to use electronic health records and machine learning to aid doctors in diagnosing patients through more efficient use of the records.
“We’re thinking of building an intelligent assistant for clinicians. Let’s say a new person comes in, and a physician tries to prescribe something,” Natarajan says. “You can have a pop-up in the electronic health record that says, ‘By the way, did you know he had this particular complaint seven years back?’’
It’s also an intelligent assistant that isn’t forgetful.
“(Physicians) have a limited amount of resources,” Natarajan says. “They’re seeing maybe 20 or 25 patients a day, and they have to recall everything, and the patients aren’t giving them all the information all the time. People forget a lot of things, but a computer doesn’t. You can pull up a record and show them what they need to know.”
Natarajan is working on developing new algorithms that will combine the medical knowledge gleaned through the centuries with the real-life data that already exists to find correlations between certain conditions.
“One electronic health record that we work with is from the 1960s,” Natarajan says. “They were one of the earliest adopters of electronic health records. They have 40 years worth of data. That’s fantastic for us to use because we can look at something that happened 20 years past to say whether it is going to have an effect right now, and that’s very, very useful. There is clinical study data that I’m working with that is nearly 25 or 30 years old. We can see that there is a decision that was made in 1985 that can possibly impact something in 2010. You can actually find these correlations.”
His dream is to see the sheer amount of data available overwhelm such variables as race, gender, and age when it comes to diagnosing conditions and predicting the long-term impact of a given treatment.
The intelligent assistant also could help physicians keep up with the latest medical information at a higher level than is possible by humans. Using natural language processing techniques, Natarajan hopes to keep physicians more informed about the impact of their treatments.
Using advanced computing techniques and electronic health records, Natarajan hopes he can improve success rates for both physicians and patients alike, something that has doctors excited about the direction of medicine.
“All the physicians that I’ve contacted have loved the concept of using advanced computing techniques and advance mathematical models to make these predictions,” Natarajan says. “They have built their models in their head when they’re diagnosing their disease. What we’re trying to do is formalize that by building a machine to help them.”