Exploring the feasibility of modeling next-day fatigue and sleepiness using digital sleep tracker data in neurodegenerative and immune-mediated inflammatory diseases

Engineering notes

Key topics: autonomous driving, prediction. See the paper for implementation details and experimental results.

Chinese explanation / 中文解读

中文解读待补充：本站会优先为端到端自动驾驶、BEV感知、3D目标检测、轨迹预测、路径规划、LiDAR感知等高价值论文补充中文说明。

Original abstract

Background Fatigue and sleep disturbances are highly prevalent in neurodegenerative diseases (NDDs) and immune-mediated inflammatory diseases (IMIDs). Conventional patient-reported outcomes (PROs) are subjective and prone to recall bias; Digital health technologies and wearable sleep trackers offer objective, continuous monitoring of sleep and physiology at home. Objective This study evaluated the feasibility of using consumer- and research- grade sleep trackers to predict next-day physical and mental fatigue and daytime sleepiness in individuals with NDDs and IMIDs as an exploratory analysis, and examined whether machine-learning models could identify preliminary sleep features to inform future fatigue monitoring research in chronic disease populations. Methods The IDEA-FAST feasibility study enrolled 134 participants (42 healthy adults, 39 NDD, 53 IMID) across four European centres. Over 3,062 nights, participants wore three sleep trackers (BedSensor, ZKONE, DREEM 2) and completed daily fatigue and sleepiness PROs at home. A polysomnography sub-study ( <mml:math xmlns:mml="http://www.w3.org/1998/Math/MathML" id="IM1"> <mml:mi>n</mml:mi> <mml:mo>=</mml:mo> <mml:mn>28</mml:mn> </mml:math> ) validated tracker performance. Machine learning models using physiological and sleep-architecture features were evaluated with leave-one-subject-out cross-validation. Results Sleep trackers showed moderate PSG agreement. Models demonstrated preliminary discriminative capacity for next-day physical fatigue in healthy adults (AUC <mml:math xmlns:mml="http://www.w3.org/1998/Math/MathML" id="IM2"> <mml:mo>=</mml:mo> </mml:math> 0.75), driven mainly by respiratory rate and REM sleep duration. In NDD, physical fatigue AUC reached 0.62 under enriched training, with REM latency and deep sleep as key features. Mental fatigue prediction reached AUC <mml:math xmlns:mml="http://www.w3.org/1998/Math/MathML" id="IM3"> <mml:mo>=</mml:mo> </mml:math> 0.66 in healthy adults; daytime sleepiness AUC <mml:math xmlns:mml="http://www.w3.org/1998/Math/MathML" id="IM4"> <mml:mo>=</mml:mo> </mml:math> 0.66 in NDD. Findings should be interpreted as exploratory, as outcome binarisation using a global threshold may conflate between-person disease-group differences with within-person symptom variation. Conclusions Wearable sleep trackers show feasibility for objective home-based sleep monitoring, with preliminary evidence supporting sleep physiology as a candidate predictor of next-day physical fatigue in healthy adults. Predictive performance in chronic disease cohorts remains limited, underscoring the need for larger, multimodal studies to establish disease-specific digital fatigue endpoints.

Links and sources

• PDF from original source

Comments