Main Content

Quantifying symptom severity in people with Parkinson's disease using kinetic and electromyographic parameters with machine learning - a pilot study

Description

In this study, patients diagnosed with idiopathic Parkinson's disease were assessed for the severity of their motor symptoms. The aim was to develop an objective method of quantifying and assessing motor difficulties as an alternative to the current subjective clinical assessment. To achieve this, kinetic and electromyographic parameters were measured in 45 subjects using a special wristband during various movement exercises. The data were then used to train selected regression models as machine learning techniques. The models were designed to predict a clinical score (MDS-UPDRS Part III) based on electromyographic parameters typically obtained during clinical assessments. The results showed high correlations between predicted and clinically measured scores in selected regression models. Another study, carried out in collaboration with the Justus Liebig University in Giessen, Germany, aimed to differentiate the levodopa-induced "ON" and "OFF" states of subjects on the basis of kinetic data. The use of both a regression model and a convolutional neural network resulted in high accuracy.

The results of our study provide a promising basis for further work and the development of a convenient system for continuous monitoring of motor symptom status in people with Parkinson's disease. Such a system could make a significant contribution to the prevention of complications and the individualisation of therapies. 

Publications:

[1] Tabatabaei, S. A. H. and Pedrosa, D.J., Eggers, C., Wullstein, M., Kleinholdermann, U., Fischer, P., & Sohrabi, K. (2020). Machine Learning Techniques for Parkinson’s Disease Detection using Wearables during a Timed-up-and-Go-Test. Current Directions in Biomedical Engineering, 6(3), 376-379.

[2] Kleinholdermann, U., Wullstein, M., & Pedrosa, D. J. (2021). Prediction of motor Unified Parkinson's Disease Rating Scale scores in patients with Parkinson’s disease using surface electromyography. Clinical Neurophysiology, 132(7), 1708-1713.

Contact

Dr. rer nat. Urs Kleinholdermann
Mr. Maximilian Wullstein
Telefon: 06421/58 - 65299
Telefax: 06421/58 - 67055
kleinhol@staff.*

* please add "uni-marburg.de" for a full email-address.