Accurate Step Count with Generalized and Personalized Deep Learning on Accelerometer Data.

Physical activity (PA) is globally recognized as a pillar of general health. Step count, as one measure of PA, is a well known predictor of long-term morbidity and mortality. Despite its popularity in consumer devices, a lack of methodological standards and clinical validation remains a major impediment to step count being accepted as a valid clinical endpoint. Previous works have mainly focused on device-specific step-count algorithms and often employ sensor modalities that may not be widely available. This may limit step-count suitability in clinical scenarios. In this paper, we trained neural network models on publicly available data and tested on an independent cohort using two approaches: generalization and personalization. Specifically, we trained neural networks on accelerometer signals from one device and either directly applied them or adapted them individually to accelerometer data obtained from a separate subject cohort wearing multiple distinct devices. The best models exhibited highly accurate step-count estimates for both the generalization (96-99%) and personalization (98-99%) approaches. The results demonstrate that it is possible to develop device-agnostic, accelerometer-only algorithms that provide highly accurate step counts, positioning step count as a reliable mobility endpoint and a strong candidate for clinical validation.

Assessment of Objectively Measured Physical Activity as an Independent Estimator of Functional Status in Clinical Trials.

Functional status of patients is an important concept in clinical trials. It subsumes functional capacity, which is traditionally estimated by exercise tests, and functional performance, which is often estimated by questionnaires. Objectively measured physical activity by means of wearables devices containing accelerometers (PA) have recently been proposed as a novel and advantageous way to estimate physical status including capacity and performance. There is nonetheless insufficient evidence of the association between PA and traditional ways to estimate functional status. In the ACTIVATE clinical trial, cycle ergometry tests were performed multiple times in all 267 patients, PA was measured for a week prior to each cycle ergometry test, and questionnaires were answered daily during the same week. Pearson's correlation tests and clustering analysis revealed that PA, physical activity experience as assessed by questionnaires, and exercise endurance time as measured by the cycle ergometry test, are largely independent. Therefore, all three approaches together might achieve a complete assessment of the functional status of patients in clinical trials, as they each independently correlate with health-related quality of life and important clinical outcomes such as hospitalizations but are weakly associated among themselves.