Consumer wearable devices can be key tools in predicting the onset of illnesses like COVID-19 by using health metrics like breathing rate, resting heart rate and heart rate variability (HRV), according to findings published in npj Digital Medicine from Fitbit.

The paper outlines predictors for the severity of illness, symptom prevalence, illness duration among male and female participants and estimations of the need for hospitalization.

TOPLINE DATA

Of the 2,745 subjects diagnosed with COVID-19 used in the study, 11.9% of males and 11.2% of females were asymptomatic; 48.3% of males and 47.8% of females recovered at home, by themselves; 29.7% of males and 33.7% of females recovered at home, with the help of someone else; 9.3% of males and 6.6% of females required hospitalization without ventilation; and 0.5% of males and 0.4% of females required ventilation.

There were 21 different symptoms reported in the study, and the prevalence of symptoms varied between the male and female participants.

Fatigue was the most common symptom, present in 66.1% of males and 76.1% of females. Headaches were the next most prevalent symptom, with 53.7% of males and 71.4% of females reporting this. Body aches followed and were present in 58.4% of males and 64.4% of females. Loss of taste and smell was reported in 50.2% of males and 64.9% of females. Rounding out the list was cough, which was reported by 55.2% of males and 59.9% of females.

Fevers were only present in 59.4% of males and 52% of females, indicating that temperature screening alone may not be enough to measure infection rates, according to the study.

The duration of symptoms depended on the severity of the illness. Mild cases showed an average duration of 11 days. Moderate cases typically lasted for 16 days. Cases that required hospitalization averaged 25 days, although the spread was large for severe cases.

The percentage of participants who experienced symptoms for more than 60 days with mild, moderate and severe cases was 3.9%, 6.4% and 16%, respectively.

Shortness of breath was found to be highly indicative of the need for hospitalization, while sore throat and stomach ache were the least likely. Gastrointestinal symptoms, such as vomiting and loss of appetite, were indicative of severe illness. Among demographic information, being older and having a high BMI showed higher likelihoods for hospitalization.

Researchers also created a classifier to predict illness on any specific day, given health metrics.

It was able to predict 21% of cases the day before symptoms began, 31% of cases the day symptoms arose and 43% of cases the day following with 90% specificity. When set to 95% specificity, it detected 15% of cases the day before symptoms began showing, 24% on the first day of symptoms and 33% of cases the day after. At 99% specificity, 6.8% of cases could be detected one day before the onset of symptoms, 12% of cases the day symptoms begin and 20% the day after.

METHODS

Data on 2,745 subjects diagnosed with COVID-19 were collected from May 21 to September 11, 2020, consisting of PCR-positive tests conducted between February 16 and September 9.

The study included active Fitbit users in the U.S. and Canada that had COVID-19 or similar infections. They were invited to participate in a survey to report whether they had been tested and what their symptoms were. Participants had the option to include demographic data like age, sex, BMI and relevant background medical information such as conditions including diabetes, coronary artery disease or hypertension.

Researchers also collected health metric data for breathing rate, resting heart rate and HRV using the participants’ Fitbit devices.

The study noted several limitations, including that the participants were all Fitbit users and may not represent the general population. They also self-selected to participate in the study and self-reported their symptoms.

“Nevertheless, we believe this survey provides an important scientific contribution by suggesting (a) hospitalization risk can be calculated from self-reported symptoms, and (b) relevant and predictive physiological signs related to COVID-19 may be detected by consumer wearable devices,” the researchers said in the paper.

THE LARGER TREND

Fitbit released early findings from this study in August, and most notably shared that its devices could detect nearly 50% of COVID-19 cases one day before participants report the onset of symptoms, with 70% specificity.

The U.S. Department of Defense recognized Fitbit’s work in early COVID-19 detection and gave it nearly $2.5 million to expand its research through a study with Northwell Health’s Feinstein Institutes for Medical Research.

This has been a big year for Fitbit research even beyond COVID-19.

The Taiwan-based diabetes management app Health2Sync announced that combining Fitbit wearable devices with their Health2Sync Patient Management Platform can help users control and better manage the symptoms of Type 2 diabetes mellitus.

Fitbit also recently released findings from a study that used its smartwatches to show how devices continuously gathering data from the sympathetic and the parasympathetic nervous systems could derive and describe diverse measures of cardiac autonomic function among users.