Detecting Parkinson's Disease through Gait Measures Using Machine Learning.

Area of Science:

• Neurology
• Biomedical Engineering
• Data Science

Background:

• Parkinson's disease (PD) is a common progressive neurodegenerative disorder affecting motor function.
• Diagnosis relies on motor signs like bradykinesia, rigidity, tremor, and postural instability, with current criteria still evolving.
• Accurate and cost-effective diagnostic methods are crucial for managing PD.

Purpose of the Study:

• To develop a machine learning classifier using gait data to distinguish between Parkinson's patients and healthy individuals.
• To explore the potential of gait analysis as a diagnostic tool for Parkinson's disease.

Main Methods:

• Utilized the Gait in Parkinson's Disease dataset from PhysioNet, including force sensor gait data from 214 PD patients and 92 healthy controls.
• Applied various machine learning algorithms: logistic regression, Support Vector Machine (SVM), decision tree, and K-Nearest Neighbors (KNN).
• Compared classifier performance against baseline models, including frequency domain methods.

Main Results:

• Machine learning models demonstrated the ability to classify Parkinson's disease patients and healthy controls based on gait data.
• Several tested algorithms achieved significant classification performance.
• Baseline frequency domain methods showed comparable performance, indicating their utility in PD diagnostics.

Conclusions:

• Machine learning analysis of gait data provides a viable method for Parkinson's disease diagnosis.
• This approach has the potential to be more accurate and cost-effective than current diagnostic methods.
• Gait analysis, particularly using machine learning, represents a promising avenue for future PD diagnostic strategies.