Comparative analysis of trunk sensor placement in people with and without Parkinson's disease

Wearable devices are widely adopted for monitoring motor symptoms in Parkinson's Disease (PD), but optimal sensor placement remains under debate. This study compares kinematic data between a smartphone placement on the abdomen (patient-friendly) and a lumbar sensor (gold-standard) in individuals with and without PD. Five PD patients and five healthy controls wore inertial measurement units (IMUs) and consumer smart devices (smartwatch and smartphone) on their wrist and around their waist. Wrist sensors were utilised only to synchronise between the two sensor systems. The lumbar IMU and smartphone were placed at the level of the L5 vertebra and in the abdomen region, respectively. Data were collected during quiet stance with closed eyes and overground walking tasks. For both accelerometer and gyroscope signals, maximum spectral coherence was used to assess the frequency-domain correlation between the smartphone and the lumbar sensor while spectral gain measured their spatial coupling in the anteroposterior, mediolateral, and vertical planes within the 0.1-10 Hz frequency band. Analysis demonstrated a good to excellent inter-device agreement (coherence > 0.9) in both groups in most coordinates. However, coherence decreased (coherence < 0.9) for mediolateral acceleration and pitch rotation in walking. Sensor agreement was substantially lower in the quiet stance tasks particularly for anteroposterior acceleration and pitch rotation. These results were evident in both groups. Abdomen device placement provides clinically valid movement data for PD motor symptom monitoring. Yet, caution is advised for specific parameters, as abdominal movement may introduce noise, affecting measurement accuracy. This research contributes to identifying the patient-centric device positioning without compromising data, in both people with PD and healthy individuals.