Tracking the longitudinal progression of Bradykinesia in Parkinson's disease from videos using VisionMD.

Abstract

BackgroundBradykinesia, a hallmark motor symptom of Parkinson's disease, is difficult to quantify accurately using clinical scales due to coarse nature of the scales and their limited sensitivity to change. VisionMD is an open-source platform for video-based quantification of bradykinesia that complements clinical scale evaluations.ObjectivesTo evaluate whether VisionMD video-based digital measures derived from short Finger Tapping videos can detect longitudinal changes in bradykinesia.MethodsWe analyzed Finger Tapping task videos from 36 individuals with Parkinson's disease at two times points, separated by an average of 1.6 years. Motor function was assessed using the Finger Tapping clinical score and video-based digital measures provided by VisionMD which capture movement speed and timing. Longitudinal changes were assessed using ordinal linear models for clinical scores and linear mixed-effects models for digital measures. Correlations between clinical scores and digital measures were also examined.ResultsDigital measures were significantly correlated to Finger Tapping clinical scores. While overall motor function worsened over time, these changes were not reflected by the Finger Tapping score. In contrast, digital measures derived from Finger Tapping videos detected longitudinal declines in movement speed, revealing longitudinal changes in bradykinesia. When stratified by baseline severity, distinct patterns emerged, highlighting heterogeneity in bradykinesia progression in PD.ConclusionVisionMD's video-based digital measures captured subtle, domain-specific progression of bradykinesia not detected by clinical scores. VisionMD provides an objective tool to monitor PD progression and disentangle the multidimensional components of bradykinesia. Future studies should focus on validating these findings in larger cohorts and from home-based recordings.