Digital Health (May 2023)

Kinect-based objective evaluation of bradykinesia in patients with Parkinson's disease

  • Zhuang Wu,
  • Hongkai Gu,
  • Ronghua Hong,
  • Ziwen Xing,
  • Zhuoyu Zhang,
  • Kangwen Peng,
  • Yijing He,
  • Ludi Xie,
  • Jingxing Zhang,
  • Yichen Gao,
  • Yue Jin,
  • Xiaoyun Su,
  • Hongping Zhi,
  • Qiang Guan,
  • Lizhen Pan,
  • Lingjing Jin

DOI
https://doi.org/10.1177/20552076231176653
Journal volume & issue
Vol. 9

Abstract

Read online

Objective To quantify bradykinesia in Parkinson's disease (PD) with a Kinect depth camera-based motion analysis system and to compare PD and healthy control (HC) subjects. Methods Fifty PD patients and twenty-five HCs were recruited. The Movement Disorder Society-Sponsored Revision of the Unified Parkinson's Disease Rating Scale part III (MDS-UPDRS III) was used to evaluate the motor symptoms of PD. Kinematic features of five bradykinesia-related motor tasks were collected using Kinect depth camera. Then, kinematic features were correlated with the clinical scales and compared between groups. Results Significant correlations were found between kinematic features and clinical scales ( P < 0.05). Compared with HCs, PD patients exhibited a significant decrease in the frequency of finger tapping ( P < 0.001), hand movement ( P < 0.001), hand pronation-supination movements ( P = 0.005), and leg agility ( P = 0.003). Meanwhile, PD patients had a significant decrease in the speed of hand movements ( P = 0.003) and toe tapping ( P < 0.001) compared with HCs. Several kinematic features exhibited potential diagnostic value in distinguishing PD from HCs with area under the curve (AUC) ranging from 0.684–0.894 ( P < 0.05). Furthermore, the combination of motor tasks exhibited the best diagnostic value with the highest AUC of 0.955 (95% CI = 0.913–0.997, P < 0.001). Conclusion The Kinect-based motion analysis system can be applied to evaluate bradykinesia in PD. Kinematic features can be used to differentiate PD patients from HCs and combining kinematic features from different motor tasks can significantly improve the diagnostic value.