Activity-independent detection of mediation states in individuals with Parkinson’s disease using wearable sensors (P2.8-004)

Objective: We investigated whether wearable sensors can be used to provide objective measurements regarding the medication states of patients with Parkinson’s disease (PD) during their activities of daily living.

Background: Motor fluctuations between akinesia (medication OFF) and mobile (medication ON) phases are one of the most prevalent complications of patients with Parkinson’s disease (PD), occurring in 50% of patients within 3 to 5 years of diagnosis and 80% of patients after 10 years. There is a need for a technology-based system that provides reliable and objective measurements of a patient’s duration medication OFF and ON phases that can be readily used by the clinician to make successful therapeutic adjustments.

Design/Methods: 19 fluctuating PD subjects (age: 42–77, 14 males) were selected in this study. Two KinetiSense motion sensors were mounted on the most affected wrist and ankle to collect gyroscope signals. The participants performed seven daily living activities in their medication OFF and ON phases: resting, walking, drinking, dressing, hair brushing, unpacking groceries, and cutting food. A series of significant features were extracted from the motion signal. A medication OFF/ON classifier based on support vector machine with fuzzy labeling was trained using approximately 15% of four activities (ambulation, drinking, arm resting, and dressing) and tested on the remaining data.

Results: The trained classifier was able to detect medication states with 90.5% classification accuracy, 94.2% sensitivity, and 85.4% specificity. The average accuracy for the activities in the training phase was 91.3%, and for the new activities it was 88.4%, indicating that the algorithm is activity-independent and performs equally-well for all the seven activities.

Conclusions: Our algorithm would enable an easy-to-use sensor-based assessment system that is not task-based, which can provide accurate and timely information to address motor fluctuations. Use of such sensor data could considerably improve both the care delivery and quality of life for PD patients.

Disclosure: Dr. Hssayeni has nothing to disclose. Dr. Burack has nothing to disclose. Dr. Jimenez Shahed has received personal compensation for consulting, serving on a scientific advisory board, speaking, or other activities with St. Jude Medical/Abbott, Medtronic, Teva, Bracket, Nuvelution, Sunovion,. Dr. Jimenez Shahed has received research support from Biotie/Accorda, Medtronic, St. Jude Medical/Abbott, Eli Lilly & Company, Wilsons Therapeutics. Dr. Ghoraani has nothing to disclose.