Leveraging Smart Bed Technology to Detect COVID-19 Symptoms: Case Study.

Area of Science:

• Digital Health
• Biomedical Engineering
• Infectious Disease Surveillance

Background:

• Viral infections like COVID-19 impact sleep and cardiorespiratory function.
• Consumer smart bed technology enables unobtrusive, real-world monitoring of sleep and physiological signals.
• Smart beds offer an alternative to wearables and self-reports for objective data capture.

Purpose of the Study:

• To use smart bed ballistocardiography (BCG) signals and predictive modeling for detecting and monitoring COVID-19 symptoms.
• To leverage objective, longitudinal biometric data for individual-level illness tracking.

Main Methods:

• Retrospective analysis of 1725 US adults using smart bed data (pulse rate, respiratory rate, sleep metrics).
• A two-stage machine learning pipeline: symptom detection and illness-symptom progression models.
• Within-subject comparisons of symptomatic versus baseline periods using Gaussian Mixture Hidden Markov Models.

Main Results:

• The model detected COVID-19 symptoms in 104 out of 122 positive cases.
• Significant deviations in sleep and cardiorespiratory metrics were observed during symptomatic periods (AUC=0.80).
• The model demonstrated high discriminatory performance in identifying illness windows.

Conclusions:

• Smart beds provide a valuable resource for passive, objective, longitudinal data collection.
• Findings support the feasibility of using smart bed data and ML for real-time COVID-19 detection.
• Future work includes model refinement and application for population-level infectious disease surveillance.