NeuroRPM

NeuroRPM is a software application for the Apple Watch that utilizes accelerometer and gyroscope data processed by machine learning algorithms to quantify the presence or absence of tremor, bradykinesia, and dyskinesia in adults with Parkinson's disease. The system aims to provide objective, frequent symptom data from both clinic and home environments, with its performance in detecting these symptoms validated through clinical testing against established neurological scales and expert assessment.

NeuroRPM, developed by New Touch Digital Inc., is a software application for the Apple Watch designed to quantify motor symptoms of Parkinson's disease in adult patients aged 46 to 85. Its primary function is to monitor and classify tremor, bradykinesia, and dyskinesia during wake periods in both clinic and home settings. The system utilizes the Apple Watch's accelerometer and gyroscope to collect motion data, which is then transmitted to cloud servers. These data are analyzed by proprietary machine learning models to generate binary classifications (presence or absence) for each symptom type every 15 minutes, referencing validated scales like the Unified Parkinson Disease Rating Scale (UPDRS) and the Abnormal Involuntary Movement Scale (AIMS) for its output interpretation. The performance of NeuroRPM was evaluated in an observational, non-interventional clinical study involving 36 subjects diagnosed with Parkinson's disease, establishing its sensitivity and specificity. Ground truth for the study was derived from scores assigned by an expert panel of three board-certified movement disorder specialists using the UPDRS and AIMS clinical scales. For tremor, NeuroRPM demonstrated a sensitivity of approximately 71.8% and specificity of 95.1%; for bradykinesia, sensitivity was 71.4% and specificity 77.4%; and for dyskinesia, sensitivity was 71.2% and specificity 94.7%. While primarily validated in a clinical setting, supportive data from subjects performing common at-home tasks were included to suggest applicability in home environments, noting the study cohort was 95.5% Caucasian.

AI Regulatory Concerns

Powered by

Jev reads the full FDA letter and returns calibrated concern probabilities per regulatory category. These are AI estimates, not regulatory advice.

• Q1. How does the NeuroRPM algorithm differentiate between Parkinsonian symptoms and other voluntary or involuntary movements common in daily life, especially in an unconstrained home environment?; Given the 15-minute interval for binary symptom classification, how does the system handle fluctuations in symptom severity within that window, and what is the minimum duration or intensity of a symptom required for detection?; The study population was 95.5% Caucasian; how might the device's performance vary across more diverse ethnic populations, and are there plans for validation in such groups?; What is the long-term reliability and consistency of the Apple Watch sensors for collecting motion data specifically for Parkinson's symptom quantification, and how are sensor drift or degradation addressed by the NeuroRPM system?; How does the binary output (symptom present/absent) compare to more granular scoring systems in terms of tracking subtle changes in response to medication or disease progression, and what is the clinical utility of this binary classification for fine-tuning treatment adjustments?