The full text of this item is not available at this time because the student has placed this item under an embargo for a period of time. The Libraries are not authorized to provide a copy of this work during the embargo period, even for Texas A&M users with NetID.
Leveraging Robust Personalized Baselining for Physiological Event Detection Using Commercial Wearables
Abstract
Data from commercial off-the-shelf (COTS) wearables combined with machine learning algorithms provide an unprecedented potential for early detection of adverse physiological events. However, several challenges dampen these potentials, including (1) heterogeneity among and within participants that poses challenges when scaling detection algorithms to a general population, (2) undesired confounders that lead to incorrect assumptions regarding a participant’s healthy state, and (3) noise in the data at the sensor level that limits the sensitivity of detection algorithms. To address these challenges, we propose a novel baseline correction algorithm leveraging deep learning techniques that build on determining suitable personalized baselines for each participant to capture important physiological changes. Our work is validated on potentially one of the largest wearable datasets available, obtained with 8000+ participants and 1.3+ million hours of wearable data captured from Oura smart rings. Leveraging this extensive dataset, we achieve prediction of COVID-19 with a performance ROC AUC of up to 0.762 with the baseline correction technique applied to the data, from 0.723 when this technique is not applied. Similarly, we achieve ROC AUCs of up to 0.868, from 0.858 for detection of fever, and up to 0.617, from 0.607 for detection of shortness of breath using Oura rings, respectively. These techniques offer improvements across most metrics and events, including PR AUC, sensitivity at 75% specificity, and precision at 75% recall. Furthermore, we demonstrate improvements in the early detection of COVID, achieving an average accurate prediction of 3.8 days before a reported positive test, an improvement over 3.5 days.
Subject
COTS WearablesCOVID-19
Physiological Event Detection
Fever
Shortness of Breath
Wearables
Health
Citation
Passage, Bryant Victor (2023). Leveraging Robust Personalized Baselining for Physiological Event Detection Using Commercial Wearables. Master's thesis, Texas A&M University. Available electronically from https : / /hdl .handle .net /1969 .1 /199106.