System uses sensors, machine learning to detect Parkinson’s

Researchers have developed a simpler way to diagnose Parkinson’s disease by combining wearable sensors and machine learning, according to a study published in Sensors. Current diagnosis methods require detailed clinical evaluations, but this new approach uses just one sensor placed on the lower back to track movement, and a streamlined machine learning process to identify patterns. In tests with over 300 participants, this method showed over 92% accuracy in distinguishing people with Parkinson’s from those without. The researchers also found that using data from a single mobility task was nearly as accurate as using data from multiple tasks. While the results are promising, further testing is needed before it can be applied in clinical practice.