**Background:** Parkinson's disease (PD) is a neurodegenerative disorder with motor symptoms including bradykinesia, rigidity, tremor, and gait impairment, as well as complications like levodopa-induced dyskinesia (LiD) and freezing of gait (FoG). Optimal management requires accurate, longitudinal symptom monitoring, but standard clinical assessments are limited to brief, episodic evaluations that may not capture day-to-day fluctuations. Wearable devices offer a potential solution for continuous, objective home monitoring, but require rigorous validation. The PDMonitor® system (PD Neurotechnology Ltd.) consists of five wearable inertial measurement unit (IMU) sensors worn on the wrists, ankles, and waist, designed to detect and quantify the full spectrum of PD motor symptoms.
**Methods:** Two multi-site clinical studies were conducted. PDNST001 enrolled 65 PD patients (mean age 65.8 ± 9 years, 33 male/30 female, mean disease duration 8.8 ± 4.9 years) across three European sites. Phase I involved 2–6 hours of in-clinic monitoring with expert UPDRS Part III and AIMS evaluations every 30 minutes. Phase II involved 1–7 days of home monitoring. PDNST002 enrolled 28 healthy controls (mean age 63.2 ± 9.9 years, 10 male/19 female) who wore the device for up to 3 days. The PDMonitor® uses machine learning algorithms to process raw IMU data (59.5 Hz sampling) into 30-minute symptom severity scores. Statistical analysis included ROC curves, confusion matrices, Bland-Altman analysis, and intra-class correlation coefficients (ICC).
**Key Results:** The PDMonitor® showed high accuracy in detecting PD motor symptoms compared to expert evaluations. For arm bradykinesia (UPDRS items 23+24+25 > 4), accuracy was 0.85 (sensitivity 0.84, specificity 0.85) with a correlation of r = 0.68. Gait impairment detection (UPDRS item 29 > 1) achieved accuracy 0.99 (sensitivity 0.67, specificity 1.0). Wrist tremor detection (UPDRS item 20 > 1) showed accuracy 0.99 (sensitivity 0.84, specificity 0.99). Leg tremor detection achieved accuracy 0.99 (sensitivity 0.93, specificity 0.99). Dyskinesia detection (AIMS > 4) showed accuracy 0.99 (sensitivity 0.82, specificity 0.99). OFF state detection compared to diaries showed accuracy 0.96 (sensitivity 0.85, specificity 0.97). FoG discrimination accuracy was 0.96 (sensitivity 0.83, specificity 0.98). For total time estimation, correlation with expert assessments was r² = 0.75 for OFF time and r² = 0.63 for dyskinesia time. Day-to-day ICCs were high: bradykinesia ICC = 0.77 (Spearman's rho = 0.83), dyskinesia ICC = 0.82 (Spearman's rho = 0.45), gait ICC = 0.71 (Spearman's rho = 0.83). Wearability assessment showed patients required an average of 5.3 ± 2 minutes to put on all five sensors. Comfort Rating Scale scores were low across all domains (most < 3/20), indicating good tolerability.
**Clinical Implications:** The PDMonitor® provides comprehensive, objective, and longitudinal monitoring of the majority of PD motor symptoms with high accuracy and good test-retest reliability. Its ability to detect OFF time (r² = 0.75) and dyskinesia time (r² = 0.63) is particularly valuable for therapeutic decision-making, as these are key targets for medication adjustment and advanced therapy selection (e.g., deep brain stimulation). The system's high specificity across all symptoms (>0.97 for most) minimizes false positives during daily activities, which is critical for home use. The device addresses a significant gap in PD management by capturing symptom fluctuations that are often missed in clinic-based assessments. Limitations include potential underestimation of dyskinesia in head/neck regions and the need for further real-world evidence. The study was funded by PD Neurotechnology Ltd., and several authors are employees or consultants of the company.