ML Tracks Huntington's Disease Progression via Smartwatch Passive Monitoring — Clinic-Free Biomarker Across 1,025 Patients

A study published in npj Digital Medicine (Nature group) on April 25, 2026 demonstrated that machine learning models analyzing passive wrist accelerometer data from smartwatches can continuously and objectively track the progression of chorea — the involuntary movement disorder that defines symptomatic Huntington's disease — over a two-year period without requiring clinic visits. The study drew on four cohorts totaling 1,025 participants across HD progression stages, training ML models on real-world daily-life accelerometry data to extract movement signatures characterizing chorea. The model detected a 0.13-point annual progression rate in upper-limb chorea using the established Total Motor Score scale — providing a continuous, sensitive biomarker that traditional in-clinic movement rating scales (conducted sporadically every 6-12 months) cannot match. Continuous passive monitoring addresses two major bottlenecks in HD research: clinic-based assessments are episodic, expensive, and cognitively burdensome for patients; and subjective inter-rater variability in movement rating reduces clinical trial sensitivity. The tool has immediate implications for HD clinical trials, where demonstrating slowing of disease progression with experimental therapies requires sensitive outcome measures. It also opens a pathway to longitudinal monitoring in settings where specialist HD clinics are unavailable — extending high-quality outcome measurement to patients in rural and lower-resource health systems. Huntington's disease is a rare but fatal genetic condition affecting approximately 41,000 people in the US and 30,000 in Europe, with no disease-modifying treatment currently approved.