Decision-making computer could redefine medicine

The future of medical technology might come in the form of a decision-making computer.

An artificial intelligence could search an entire patient’s medical history, warn of complications and recommend possible treatments based on a patient’s current stage of illness. Such a computer is the pet project of two IU computer science and informatics professors, Kris Hauser and Sriraam Natarajan.

They, along with Dr. Shaun Grannis, a biomedical informatics physician from the Regenstrief Institute, have been awarded a National Science Foundation exploratory grant for more than $686,000 to pursue the development of artificial intelligence in medicine.     

Still in its early stages, the project uses statistical models from clinical studies, medical data and patient histories to analyze a patient’s vital information and symptoms to provide doctors with a list of possible treatments.

“A doctor uses information that is readily available and a short screening to then plan a treatment based on what he or she has inferred,” Natajaran said. “We are looking for a way to mathematically formulate this.”

More than 40 years have passed since the first integration of computers into the patient care settings.

The computerized records of medical data have been constantly expanding ever since.

“The idea is to fuse the raw data with your entire patient history and then use the models to track relationships and provide more reliable recommendations,” Hauser
said.

The initial models being developed in this exploratory project are cardiology, clinical depression and emergency room readmission.

“After we have analyzed the medical data on these specific areas, written code and tested the system, we will have a clearer view of whether this concept can be expanded upon and put into actual use,” Natarajan said.

Hauser said the current evolution of medical technology is closer to supporting this type of technology but is incompatible due to the differing formats of medical records.

“Most medical data is being recorded on computers,” Hauser said. “The problem is that the formats aren’t the same. In the future, I think we will see a push for a standardization of formatting. Then all hospitals, physicians and insurance companies will be able to share data.”

The scientists said they acknowledge the possibility of concern about computers making medical decisions instead of a traditional doctor, but dispute the difference
between decision making and decision assistance.

“We are not manipulating the data in any way,” Natajaran said. “We are certainly not looking to replace a physician. The goal is to provide a system that can analyze a life time of medical history and data, which will then provide solutions for a doctor to consider.”