Parkinson's disease detection using olfactory loss and REM sleep disorder features

Area of Science:

• Neurology
• Biomedical Engineering
• Data Science

Background:

• Parkinson's disease (PD) has a prodromal phase with non-motor symptoms like olfactory loss and sleep disorders, preceding motor symptoms by years.
• Early diagnosis of PD is crucial for timely intervention and management.
• Machine learning models offer potential for early PD detection using premotor features.

Purpose of the Study:

• To develop and evaluate automated diagnostic models for early Parkinson's disease detection.
• To utilize olfactory loss and sleep behavior disorder features for PD diagnosis.
• To assess the efficacy of Support Vector Machine (SVM) and classification tree methods.

Main Methods:

• Utilized olfactory loss data from the 40-item University of Pennsylvania Smell Identification Test (UPSIT).
• Employed sleep behavior disorder data from the Rapid eye movement sleep Behavior Disorder Screening Questionnaire (RBDSQ).
• Developed diagnostic models using Support Vector Machine (SVM) and classification tree algorithms on data from the Parkinson's Progression Marker's Initiative (PPMI) database.

Main Results:

• The developed machine learning models demonstrated high accuracy and sensitivity in classifying individuals with potential early Parkinson's disease.
• The models effectively integrated olfactory and sleep disorder features for diagnostic purposes.
• Both SVM and classification tree methods showed promising performance.

Conclusions:

• Automated diagnostic models using UPSIT and RBDSQ features show significant potential for the early detection of Parkinson's disease.
• The use of quick, inexpensive, and self-administered questionnaires like UPSIT and RBDSQ facilitates accessible early screening.
• These findings support the integration of machine learning with premotor symptoms for improved PD diagnostics.