Video-Based Hand Movement Analysis of Parkinson Patients before and after Medication Using High-Frame-Rate Videos and...

Area of Science:

• Biomedical Engineering
• Neurology
• Digital Health

Background:

• Tremor is a primary symptom of Parkinson's disease (PD).
• Contactless monitoring of PD patient movements is advancing with digital technologies.
• Artificial intelligence (AI) offers new tools for quantitative movement analysis.

Purpose of the Study:

• To quantitatively assess hand movements in PD patients using AI-based MediaPipe hand-tracking.
• To evaluate the effectiveness of standard oral medication on tremor frequency and amplitude.
• To establish a reliable, contactless method for PD tremor analysis.

Main Methods:

• Recorded high-frame-rate videos and accelerometer data from 11 PD patients.
• Collected data in both OFF-medication and ON-medication (30 minutes post-dose) states.
• Utilized AI hand-tracking to analyze video data for movement frequency and amplitude.

Main Results:

• Video-extracted data strongly correlated with accelerometer measurements.
• AI approach accurately identified tremor frequency (mean absolute error 0.229 Hz) and amplitude.
• Medication significantly reduced mean tremor frequency (2.012 Hz to 1.526 Hz) and amplitude (8.167 a.u. to 4.033 a.u.).

Conclusions:

• Achieved automatic estimation of movement and tremor frequency with low error rates.
• Demonstrated the first automated tremor analysis before and after medication in PD patients using high-frame-rate video.
• Validated AI-based video analysis as a viable tool for PD symptom monitoring.