Computer Vision-Driven Digitalization of the Nine Hole Peg Test Assessment Method: A Pilot Study

Purpose: This study aimed to develop and validate a computer vision-driven Digitalized Nine Hole Peg Test (D-NHPT) to assess hand function in stroke patients, examining the reliability and validity of extracted hand features and their ability to distinguish stroke patients from healthy subjects. Methods: A customized data collection system and an improved test device using LMC2 captured hand-motion data. The study recruited 10 stroke patients and 5 healthy subjects. Statistical analyses included intraclass correlation coefficients (ICC) for reliability, p-values for discriminant validity (Mann-Whitney U test), and |r-scores| for convergent validity. Results: The D-NHPT demonstrated high reliability (patient group ICC = 0.818–0.946; healthy group ICC = 0.785–0.904), significant discriminant validity (p < 0.019), and strong convergent validity (|r-score|=0.671–0.909). Key features included motion speed, coordination, and task completion metrics, which effectively distinguished stroke patients from healthy subjects. Conclusion: The D-NHPT provides a reliable, valid, and multidimensional assessment of hand function in stroke patients. Specific hand features are sensitive metrics for clinical evaluation, advancing digitalization of rehabilitation scales, and supporting personalized rehabilitation strategies.