[Automatic recognition and analysis of hemiplegia gait]

Area of Science:

• Biomedical Engineering
• Rehabilitation Technology
• Gait Analysis

Background:

• Hemiplegic gait analysis is crucial for rehabilitation.
• Objective and accurate gait assessment methods are needed.
• Previous methods may lack precision or efficiency.

Purpose of the Study:

• To develop an automated system for hemiplegic gait identification.
• To extract and rank the significance of gait features.
• To provide a new reference for intelligent diagnosis of hemiplegia.

Main Methods:

• Utilized Microsoft Kinect for Windows v2 to capture walking trajectory data.
• Extracted gait features: pace, stride, and center of mass movement.
• Applied Bayesian classification for gait recognition and Random Forest for feature significance analysis.

Main Results:

• Achieved 96% accuracy in classifying hemiplegic gait using Bayesian algorithm.
• Identified step speed and stride as the most significant features.
• Determined the importance ranking: step speed, stride, center of mass (left-right), center of mass (up-down).

Conclusions:

• Automated hemiplegic gait identification is feasible with Kinect data.
• Step speed and stride are critical indicators for diagnosing hemiplegia.
• The proposed method offers a valuable tool for clinical assessment and rehabilitation.