JMIR Publications

ABSTRACT

Background:

To provide effective care for COVID-19 inpatients, clinical practitioners need systems that monitor patient health and subsequently allow for risk scoring. Existing approaches for risk scoring in COVID-19 patients focus primarily on intensive care units with specialized medical measurement devices, but not on general wards.

Objective:

In this paper, we aim to develop a risk score for COVID-19 inpatients in general wards based on consumer-grade wearables (smartwatches).

Methods:

We use consumer-grade wearables to record physiological measurements such as heart rate, heart rate variability, and respiration frequency. Based on Bayesian survival analysis, we validate the association between these measurements and the patient outcomes (i.e., discharge or intensive care unit admission). To build our risk score, we generate a low-dimensional representation of the physiological features. Subsequently, a pooled logistic least absolute shrinkage and selection operator (LASSO) regression infers the probability of either hospital discharge or intensive care unit (ICU) admission.

Results:

We evaluate the predictive performance of our developed system for risk scoring in a single-center, prospective study based on N=40 inpatients with COVID-19 in a general ward of a tertiary referral center in Switzerland. First, the Bayesian survival analysis shows that physiological measurements from consumer-grade wearables are significantly associated with the patient outcomes (i.e., discharge or intensive care unit admission). Second, our risk score achieves a time-dependent area under the receiver operating characteristic curve of 0.75 to 0.94 based on leave-one-subject-out evaluation.

Conclusions:

Our results demonstrate the effectiveness of consumer-grade wearables for risk scoring in COVID-19 inpatients. Due to their low cost and ease of use, consumer-grade wearables may enable a scalable monitoring system. Clinical Trial: The study Wearable-based COVID-19 Markers for Prediction of Clinical Trajectories (WAVE) is registered at https://clinicaltrials.gov (Identifier: NCT04357834).