3D Pose Estimation for Gait Detection in Parkinson's Disease
Author Information
Author(s): He Rong, You Zijing, Zhou Yongqiang, Chen Guilan, Diao Yanan, Jiang Xiantai, Ning Yunkun, Zhao Guoru, Liu Ying
Primary Institution: University of Hong Kong-Shenzhen Hospital
Hypothesis
Can a multi-level 3D pose estimation framework using monocular video improve gait detection in Parkinson's disease?
Conclusion
The proposed 3D pose estimation method shows high reliability and effectiveness for early prediction of Parkinson's disease.
Supporting Evidence
- The classification model achieved an accuracy of 93.3%.
- Most estimated gait parameters had an ICC greater than 0.70.
- The method is low-cost and portable, making it suitable for clinical use.
Takeaway
This study created a new way to analyze how people with Parkinson's disease walk using just a video, which can help doctors spot problems early.
Methodology
The study used a multi-level 3D pose estimation framework integrating monocular video with Transformer and Graph Convolutional Network techniques to analyze gait parameters.
Limitations
The study only assessed a limited set of gait parameters and did not explore advanced features like gait symmetry.
Participant Demographics
25 healthy elderly and 34 Parkinson's disease patients, with an average age of 69.79 years for PD patients.
Statistical Information
P-Value
p<0.05
Statistical Significance
p<0.05
Digital Object Identifier (DOI)
Want to read the original?
Access the complete publication on the publisher's website