Wearable Sensors Are Reshaping How We Monitor Children With Neurological Disorders

Traditional neurological assessments in children rely heavily on clinical scales, questionnaires, and brief clinic observations—methods that are subjective, capture only a single time point, and can be biased by patient fatigue or motivation. For rapidly progressive or fluctuating conditions like Duchenne muscular dystrophy (DMD), spinal muscular atrophy (SMA), or epilepsy, these limitations are especially consequential. Wearable sensors—placed on wrists, ankles, hips, chests, or embedded in clothing and footwear—offer a compelling alternative by continuously recording biological and movement data in real-world environments.

This narrative review by González Barral and Servais synthesized 84 studies from a PubMed search using terms such as 'wearable sensors,' 'child,' and 'neurological disorder.' The authors organized findings across 14 paediatric neurological conditions. For cerebral palsy (CP), multiple studies demonstrated that accelerometers worn at the wrist, hip, or ankle can reliably classify physical activities, quantify upper limb asymmetry, assess dystonia progression at home using personalized machine learning models, and predict clinical scale scores from acceleration data. In epilepsy, wrist-worn devices combining accelerometry, electrodermal activity (EDA), and photoplethysmography (PPG) showed promise for detecting generalized tonic-clonic and focal seizures, though performance varied by seizure type and sensor placement. For autism spectrum disorder (ASD), sensors measuring EDA, heart rate variability (HRV), and movement patterns helped identify autonomic dysregulation and repetitive behaviors, with potential applications in emotion recognition and meltdown prediction.

In neuromuscular diseases, stride velocity 95th centile (SV95C) derived from ankle-worn inertial sensors emerged as a sensitive digital biomarker for DMD and SMA, capable of detecting treatment-related improvements and disease progression even over short clinical trial windows. For SMA specifically, wrist and ankle accelerometers distinguished motor function across disease severity levels and tracked responses to therapies like nusinersen. In Rett syndrome, wrist accelerometry captured hand stereotypies and sleep disturbances central to the phenotype. Down syndrome, Angelman syndrome, and Prader-Willi syndrome studies were more preliminary but demonstrated feasibility of monitoring activity levels, sleep, and motor milestones with wearables.

The review also covers ataxia (gait sensors detecting balance deficits), Gaucher disease (ambulatory monitoring of gait quality), headache (actigraphy-based sleep and activity patterns as migraine triggers), and sleep disorders broadly (wrist actigraphy as a polysomnography alternative). Across all conditions, machine learning algorithms increasingly process raw sensor data into clinically meaningful endpoints.

Despite this progress, the authors flag critical limitations: most studies involve small samples, lack longitudinal follow-up, and have not undergone formal regulatory validation. Compliance, especially in young children or those with cognitive impairment, remains challenging. Ethical concerns around continuous data collection, privacy, and parental consent are underaddressed. Regulatory approval of digital biomarkers derived from wearables in paediatric neurology is rare. The authors conclude that while wearable sensors hold transformative potential for clinical trials and routine care, substantial validation work lies ahead.