Video processing and AI provide new way to track Parkinson's disease progression

Virtually all areas of innovation have been touched by artificial intelligence, and medical technology is no exception. In recent research at the University of Florida, a team led by Diego Guarín has developed a method that uses machine learning to analyse videos of patients performing a standard test for Parkinson's disease known as the “finger-tapping test”.

Traditionally, the test would be performed under observation from a clinician. However, the team's method can detect small details of the patient's motion during the test, which would be difficult or even impossible for the clinician to observe. This can provide additional information for the clinician assessing the patient. The team's method can also be performed outside of a clinical setting: whilst the model was trained on a supercomputer, the trained model can be run on a smartphone. This means that patients can assess themselves at home, making it much easier to track progression of the disease over time.

Protection of AI inventions is a rapidly developing area of patent law, and it will be interesting to see if the University of Florida seek patent protection for their method. I was not able to find any patent applications relating to this development, but that is unsurprising for such recent research, which was published in June 2024. Patent applications are not published for up to 18 months after they have been filed, and so we might not see any such applications for a year or more.