A comparison of machine learning classifiers for smartphone-based gait analysis.

Area of Science:

• Biomedical Engineering
• Machine Learning
• Rehabilitation Technology

Background:

• Traditional patient motion monitoring systems are costly and space-intensive, limiting home rehabilitation applications.
• Reliable, automated recording and classification of patient movements are essential for telemedicine protocols.

Purpose of the Study:

• To compare state-of-the-art machine learning classifiers for gait movement analysis using smartphone-collected data.
• To develop a robust methodology for classifying gait movements to aid in patient monitoring and differentiate pathological from physiological gaits.

Main Methods:

• Stride data were collected using smartphone sensors.
• Various machine learning classifiers were evaluated for their performance in gait classification.

Main Results:

• Smartphone-based data collection offers a cost-effective and compact solution for gait monitoring.
• The study identified effective machine learning methodologies for accurate gait classification.

Conclusions:

• Smartphone-based gait analysis is a viable and practical approach for remote patient monitoring in home-based rehabilitation.
• This technique facilitates continuous patient oversight and aids in diagnosing gait abnormalities.