New research suggests that wearable activity trackers that monitor the changes in your skin temperature, heart and breathing rates, combined with artificial intelligence, could be used to identify an infection days before the symptoms start.
The findings were based on a tracker called the Ava bracelet, which is a regulated and commercially available fertility tracker that monitors breathing rate, heart rate, heart rate variability, wrist skin temperature, blood flow and sleep quantity and quality.
Researchers wanted to see if monitoring physiological changes could help develop a machine-learning algorithm to detect COVID in people who could be spreading the infection days before they know they have the virus.
Participants were from a study started in 2010 to understand the development of cardiovascular risk factors in the European country of Liechtenstein.
For this study, published June 22 in BMJ Open, the team drew 1,163 people from the study between March 2020 and April 2021.
The participants wore the bracelet at night, then used an app to record any activities that could alter central nervous system functioning, including alcohol use, prescription medications and recreational drug usage, as well as to record possible COVID-19 symptoms. They also took regular rapid antibody tests for COVID-19 and a PCR swab if they had any symptoms suggesting the virus.
About 11% of the study group, 127 people, developed COVID-19 infection during the study period. A significantly higher proportion of those who did develop COVID said they had been in contact with people in their household or work colleagues who also had COVID.
About 52% of those COVID patients, 66 in all, had worn their bracelet for at least 29 days before the start of symptoms and were confirmed as positive by PCR swab test. Those were the individuals who were included in the final analysis.
The investigators found significant changes in all five physiological indicators during the incubation, pre-symptomatic, symptomatic and recovery periods of COVID-19, compared with baseline measurements.
The algorithm was 'trained' using 70% of the data from 10 days before the start of symptoms within a 40-day period of continuous monitoring of the 66 people who tested positive for the virus. It was then tested on the remaining 30% of the data. COVID-19 symptoms in participants lasted an average of 8.5 days.
About 73% of laboratory confirmed positive cases were picked up in the training set. About 68% were found in the test set, up to two days before the start of symptoms.
The results may not be more widely applicable, the researchers said. The sample was also small, young and not ethnically diverse. Accuracy achieved was below 80%.
The algorithm is now being tested in a much larger group of people in the Netherlands. Results from that study including 20,000 people are expected later this year.
"Our findings suggest that a wearable-informed machine learning algorithm may serve as a promising tool for pre-symptomatic or asymptomatic detection of COVID-19," said the study authors, led by Dr. Lorenz Risch, from the Dr. Risch Medical Laboratory in Vaduz, Liechtenstein.
"Wearable sensor technology is an easy-to-use, low-cost method for enabling individuals to track their health and well-being during a pandemic," the researchers said in a journal news release.
"Our research shows how these devices, partnered with artificial intelligence, can push the boundaries of personalized medicine and detect illnesses prior to [symptom occurrence], potentially reducing virus transmission in communities," they concluded.
The U.S. Centers for Disease Control and Prevention has more on COVID-19.
SOURCE: BMJ Open, news release, June 22, 2022