Machine learning models for Parkinson's disease detection and stage classification based on spatial-temporal gait...

Area of Science:

• Biomedical Engineering
• Neurology
• Data Science

Background:

• Parkinson's disease (PD) diagnosis and monitoring rely heavily on subjective clinical evaluations.
• Objective diagnostic tools are needed to improve accuracy and consistency in PD management.

Purpose of the Study:

• To investigate the efficacy of machine learning (ML) algorithms in distinguishing individuals with PD from healthy controls.
• To evaluate ML algorithms for discriminating between different stages of PD using gait parameters.

Main Methods:

• Gait data was collected from 63 individuals with PD and 63 matched healthy controls during self-selected walking.
• ML algorithms, including Naïve Bayes and Random Forest, were applied to spatial-temporal gait parameters.

Main Results:

• The Naïve Bayes algorithm achieved 84.6% accuracy in PD diagnosis, identifying step length, velocity, width, and step width variability as key features.
• The Random Forest algorithm reached an Area Under the ROC curve of 0.786 for PD stage identification, highlighting stride width variability and step double support time variability.

Conclusions:

• ML analysis of gait parameters shows significant potential for objective PD diagnosis and stage identification.
• This approach can assist clinicians in the diagnosis and follow-up of Parkinson's disease.