Wearable ring outperforms spot measurements in detecting fever

A wearable ring device that can continuously read body temperature appears powerful enough to detect a fever even before it hits, according to a new study. Such a technology may help improve control measures against the novel coronavirus disease (COVID-19).

“It is likely that our current reliance on single-point temperature assessment has led to missed case identification, as indeed, the COVID-19 pandemic has not abated despite growing use of temperature checks, for example, upon entry to restaurants, stores, and air travel,” the researchers said.

Fifty adults (mean age, 43.7±11.0 years; 66 percent male) participated in the study. To be eligible for inclusion, participants must own and use an Oura smart ring and its paired smartphone app. The wearable was able to reliably detect temperature spikes during fevers, reflecting a mean increase of 0.63±1 °C (p=0.024) when participants were symptomatic relative to baseline. [Sci Rep 2020;10:21640]

However, there was substantial intra-individual variability in measurements, suggesting that using just one temperature data point would be insufficient.

Instead, multiple and continuous readings may be more instructive. Even within the same day, the temperature of a single participant would vary considerably, displaying a large difference between that day’s minimum and maximum reading. Importantly, only those who would go on to contract fevers showed a gradual rise in their daily peak temperature concordant with symptom onset.

Looking at the changes in daily minimum and maximum readings provided clear signals of fever. While the maximum temperature rose sharply in the days leading up to fever symptom onset, the minimum readings showed a gentler slope. In both cases, though, statistical significance with respect to baseline was achieved.

To develop the trend in temperature changes into digital biomarkers, the researchers assigned thresholds to the normalized daily maximum and minimum readings. Days when measurements exceeded these thresholds were labelled as fever-like days. Such instances were detected in three of 38 participants who developed fevers and in seven of 12 who did not.

The researchers then looked at other physiological markers to better understand disagreements between the fever-like days, as detected by the digital biomarkers, and self-reported fevers.

They found that the digital biomarkers were superior. When grouping participants according to fever-like days, significant changes from baseline in terms of physiological parameters such as heart rate (p=0.02) and heart rate variability (p=0.03) were detected. When grouping according to self-reported fevers instead, no such deviations were observed.

Moreover, in the 45 days before actual fever onset, 38 of the 50 participants had experienced fever-like days before they were able to self-report symptom onset.

Nevertheless, despite a clear advantage of using maximum and minimum temperature trends over single measurements, considerable heterogeneity and variance between participants and across different episodes still existed.

“Rather than pushing the utility of these specific digital biomarkers in detecting conditions such as COVID-19, this manuscript is meant to encourage future research and applications; ideally, development of such biomarkers would be coupled to symptom report and further supported by serology or other physiological confirmations of specific conditions,” the researchers said.

“Best practices need to be developed to grow the potential of distributed, participatory, wearable-enabled research into a reality that is stable, safe, and productive for all parties,” they added.