Utilizing wearable device data for syndromic surveillance: A fever detection approach
This peer-reviewed validation study published in Sensors evaluated the feasibility of using physiological data from consumer wearables for large-scale public health screening. Analyzing a massive dataset from 16,794 participants, researchers evaluated nocturnal metrics to isolate 724 self-reported fever events against 342,430 non-fever baseline days.
Using a tree-based machine learning classifier trained on nocturnal skin temperature trends, heart rate, and sleep patterns, the model successfully identified acute fever onset. The algorithm achieved an AUROC of 0.85 and demonstrated a low false positive rate of 0.8% at 0.50 sensitivity, showcasing how objective insights from Oura Ring support live health screening.
“Implementing these models could increase our ability to detect disease prevalence and spread in real-time during infectious disease outbreaks.”
Notes