Motor Fluctuations Exploration in People with Parkinson's Disease Using ARIMA-Based Accelerometer Data Modeling

Background: Parkinson's Disease (PD) is characterized by motor fluctuations that affect quality of life and require accurate monitoring to optimize treatment. Clinical outcome measures and patient diaries often fall short in capturing real-time and objective data, thus highlighting the potential of wearable sensors for symptom monitoring. Objective: This study aimed to analyze motor fluctuations in People with PD (PwPD) using accelerometer data collected during a sequence of motor tasks, such as standing and walking, performed before and after medication intake. Methods: Accelerometer data from six female PwPD were analyzed using magnitude values from pre-medication ('OFF' state) and post-medication ('ON' state) tasks. An Autoregressive Integrated Moving Average (ARIMA) model was selected based on time series data analysis, evaluated using R-squared values and Ljung-Box statistics. Results: Selection analysis resulted in AR (1) model that captured temporal dependencies in magnitude values, indicating non-random motor fluctuations. Premedication rounds showed higher R-squared values (0.499 0.804) and less variability than post-medication rounds (0.114 0.606). Ljung-Box statistics confirmed the adequacy of the model, with p-values showing no significant autocorrelation in the residuals. Conclusions: ARIMA-based modeling of wrist-worn accelerometer data reflects motor fluctuations and the effects of levodopa in managing motor symptoms, shedding light further on its role in increased movement variability. Future research with a larger sample size and longer follow-up is needed to validate these findings and to explore the role of wearable sensors in treatment optimization.