Objective: As primary objective, we acquire a data pool of medically-treated PD patients with motor fluctuations to determine the sensitivity and specificity of algorithms capable to classify the motor state in time.

Background: Motor fluctuations in Parkinson’s disease are often unpredictable and transitions between motor states (e.g. between hypokinesia and dyskinesia) often occur in short time intervals. Continuous therapies like medication pumps and subthalamic nucleus deep brain stimulation alleviate these fluctuations, however, provide rather rigid treatment regimens. Closed-loop applications might pave the way towards more personalized regimens in future, however, critically depend on valid biomarkers.

Methods: In first pilot recordings, we captured motor transitions between motor states under ongoing clinical survey and additional video documentation allowing for accurate clinical categorization. For cross-validation, we record accelerometry and gyrometry in terms of mobile inertial sensors (APDM mobility lab). We performed first exploratory analysis on these signals based on power spectral estimates.

Results: In these first preliminary analyses, we found that dyskinesias (resting position) were paralleled by a low-frequency activity increase below 4Hz. However, we expect relevant confounding from voluntary movements under daily life conditions, and expect to refine the motor state detection by adding more sophisticated classification algorithms. To this end, we plan to validate independent component analyses, and supervised learning algorithms to render the motor state classification more robust.

Conclusions: Biomarkers for automated motor state detection shall be obtained and are believed to assist personalized therapy in future.

« Back to 2016 International Congress

MDS Abstracts - https://www.mdsabstracts.org/abstract/sensor-based-motor-state-detection-in-parkinsons-disease-to-approach-personalized-therapy-delivery/