Objective: To evaluate the use of artificial intelligence in the analysis and classification of movements in Parkinson’s disease.

Background: Artificial intelligence (AI) is a powerful tool that is now being harnessed in many ways to advance medicine.  AI can process visual data to analyze movements and make classifications.  In this study, we evaluated a novel AI algorithm designed to analyze movements in patients with Parkinson’s disease (PD).

Method: Patients with PD and healthy controls were videotaped performing the movements used for the Movement Disorder Society Unified Parkinson’s Disease Rating Scale (MDS-UPDRS) motor subscale (part III).  Six components (right/left finger tapping, right/left hand movements, right/left pronation-supination) were evaluated and graded by human raters and by the AI algorithm.  Human raters and team members executing the AI algorithm were blinded to participant diagnosis.

Results: Ten patients with PD and ten healthy controls were recruited. Fifty percent were men, and the mean age was 68.4 years.  Participants with PD had a mean disease duration of 7 years, median Hoehn and Yahr score of 2, and mean MDS-UPDRS part III score of 37.1.  The average length of decrements, as quantified by the algorithm, was 36.3 for PD and 22.6 for controls (p = 0.03).  The average number of interruptions was 7.6 for PD and 5.2 for controls (p = 0.09).  The kinematic measurements that proved to be the best discriminators between PD and healthy controls were decrements mean length (AUC 0.73), peak regression slope (AUC 0.66), halt count (AUC 0.64), regressions mean (AUC 0.64), and velocity regression slope (AUC 0.64).

Conclusion: The AI algorithm successfully evaluated visual data of patients’ movements and quantified movement parameters.  Analysis of decrement, interruptions, and velocity through the algorithm demonstrated a low-to-moderate ability to differentiate PD participants from healthy controls.  AI analysis of movement is a promising tool that may one day serve as a more objective measurement of disease progression and treatment efficacy for routine care and clinical trials.

« Back to 2024 International Congress

MDS Abstracts - https://www.mdsabstracts.org/abstract/artificial-intelligence-and-visual-analysis-of-movements-in-parkinsons-disease/