Data-Driven Models for Objective Grading Improvement of Parkinson's Disease.

Area of Science:

• Neurology
• Biomedical Engineering
• Data Science

Background:

• Parkinson's disease (PD) is a progressive neurodegenerative disorder impacting motor function.
• Objective assessment of motor symptoms is crucial for effective PD management and patient care.
• Current evaluation methods can be subjective, highlighting the need for quantitative tools.

Purpose of the Study:

• To develop and validate data-driven regression models for predicting Parkinson's disease severity.
• To investigate the utility of kinematic features derived from wearable sensors for PD assessment.
• To establish an objective, motion-based system for monitoring PD progression.

Main Methods:

• Sixty-four patients with PD (PwPD) and 50 healthy controls (HC) performed 13 motor tasks.
• Wearable inertial sensors captured kinematic data during task performance.
• Data analysis involved feature extraction, selection, and regression modeling, including adaptive neuro-fuzzy inference system (ANFIS).

Main Results:

• One hundred ninety kinematic features were extracted, with a subset identified for distinguishing between PwPD and HC.
• The ANFIS model achieved a maximum correlation coefficient of 0.814 in predicting PD severity levels.
• The developed model demonstrated significant accuracy in assessing motor performance related to PD.

Conclusions:

• Kinematic features from wearable sensors can reliably predict Parkinson's disease severity.
• The ANFIS-based predictive model shows promise as an objective decision support system for clinicians.
• This approach can enhance the monitoring of PD patients over time through objective motion analysis.