Sensor-Based Quantification of MDS-UPDRS III Subitems in Parkinson's Disease Using Machine Learning

Area of Science:

• Biomedical Engineering
• Neurology
• Wearable Technology

Background:

• Parkinson's disease (PD) significantly impacts motor function, necessitating objective quantification methods.
• Current clinical assessments for PD motor symptoms can be subjective and infrequent.
• Wearable sensors offer a promising avenue for continuous, objective motor symptom monitoring.

Purpose of the Study:

• To evaluate the efficacy of wearable inertial measurement units (IMUs) combined with machine learning for quantifying upper limb motor symptoms in Parkinson's disease.
• To classify and predict specific subitems of the Movement Disorder Society-Unified Parkinson's Disease Rating Scale (MDS-UPDRS) III using sensor data.
• To establish a foundation for remote, home-based monitoring of PD motor function and treatment responses.

Main Methods:

• Utilized two compact IMUs attached to the dorsal side of each hand in 33 PD patients and 12 controls.
• Collected sensor data during six standardized clinical movement tasks, concurrently with MDS-UPDRS III assessments.
• Trained supervised machine learning models, including Random Forest (RF) and Support Vector Machines (SVM), on sensor data and clinical scores.

Main Results:

• Achieved 94% overall accuracy in classifying distinct movement tasks.
• Demonstrated high performance in classifying motor scores, with averaged Area Under the Receiver Operating Characteristic (aROC) values ranging from 68% to 92%.
• RF regression models successfully predicted motor scores; SVM models showed superior performance for specific tasks compared to RF.

Conclusions:

• Wearable IMUs coupled with machine learning provide a robust method for objective upper limb motor symptom assessment in Parkinson's disease.
• The developed methodology surpasses existing literature benchmarks in certain aspects, offering improved accuracy and prediction.
• This approach supports the development of scalable, home-based monitoring solutions to complement clinical evaluations and track treatment efficacy.