Machine Learning Approaches in Parkinson's Disease.

Area of Science:

• Neurology
• Computer Science
• Biomedical Engineering

Background:

• Parkinson's disease is a prevalent neurodegenerative disorder.
• Current diagnosis relies heavily on clinical assessment, lacking definitive in vivo biomarkers.
• Accurate and early diagnosis remains a significant challenge.

Purpose of the Study:

• To review the application of machine learning algorithms in Parkinson's disease diagnosis and characterization.
• To explore how machine learning aids in identifying patterns for disease classification.
• To provide an overview of current machine learning techniques used in Parkinson's research.

Main Methods:

• A systematic literature search was performed on PubMed using keywords 'Machine Learning' and 'Parkinson Disease'.
• The search was conducted in December 2019, yielding 230 publications.
• A total of 166 relevant articles were analyzed after excluding non-English or unrelated papers.

Main Results:

• Publications were categorized into six key application areas: Gait Analysis, Upper Limb Motor Evaluation, Handwriting/Typing, Speech Evaluation, Neuroimaging, and Metabolomics.
• Machine learning algorithms demonstrated utility across diverse data types for Parkinson's disease assessment.
• The analysis highlighted the potential of machine learning in extracting meaningful patterns from complex datasets.

Conclusions:

• Machine learning algorithms are powerful statistical tools capable of identifying patterns in large datasets.
• These approaches can assist clinicians in classifying Parkinson's disease patients based on multiple variables simultaneously.
• Machine learning holds promise for enhancing the accuracy and efficiency of Parkinson's disease diagnosis and characterization.