Historically and even today, Parkinson's disease (PD) is a clinical diagnosis, wherein a physician observes signs of the disease and an individual reports symptoms. In routine practice, there is no blood test, other biomarker, or machine to make the diagnosis or to track progression of the disease. But given advances in computing power and through computer analysis of massive amounts of data, artificial intelligence (AI) may add a valuable tool to the diagnostic process. In one form of AI, a computer analyzes a stream of input data to discern patterns that represent an outcome of interest.

A recent study used AI to non-invasively collect and analyze data on breathing patterns, using one night of breathing signals from 7,671 individuals with PD as they slept. One of the co-authors of the study was Aleksander Videnovic, MD, MS of Harvard Medical School and Massachusetts General Hospital, where he is Chief of the Division of Sleep Medicine. The hospital is a Parkinson's Foundation Center of Excellence. In this episode, he explains how the study was done, its findings, and how AI may be useful for diagnosis of PD, gauging its severity, following its progression, and possibly, in the future, assessing risk of PD before its clinical diagnosis.