Machine-based automated rating in Parkinson’s disease


  • Machine learning-based algorithms that automatically rate two cardinal symptoms of Parkinson’s disease (PD), resting tremor and bradykinesia, are feasible.

Why this matters

  • Machine learning techniques show promising imaging and video analysis capabilities in clinical settings, including quantification of human movement; however, there are currently few studies assessing vision- and machine learning-based automatic analysis of PD.

  • The reliability of machine-based automated rating in PD is a promising addition to the current remote assessment arsenal for PD management; however, improvement of current technical errors and further study for validation is required.