(Reuters Health) – Biometric sensors worn on the wrist were able to detect flu and cold viruses before people became symptomatic and predict the severity of infection, a small pilot study found.
Analysis of data from 63 volunteers who wore wrist bands with biometric sensors and were challenged with either the H1N1 flu virus or a common cold rhinovirus, found the devices could detect who was or was not infected with accuracy around 90%, and could also distinguish between mild and moderate infections 24 hours before symptom onset, according to the results published in JAMA Network Open.
“The overall take-home message is that smart watches can be used to detect infections,” said study coauthor, Jessilyn Dunn, an assistant professor of biomedical engineering and biostatistics and bioinformatics at Duke University in Durham, NC. “More specifically, they can be used to detect who might be infected without knowing it themselves and letting them know they should get a diagnostic test.”
The sensors detect changes in resting heart rate, heart rate variability, temperature, movement patterns, electrodermal skin activity and skin temperature. Dunn and her team built models that would take those inputs and calculate the likelihood the wearer was infected.
The researchers used data from two separate challenge studies, one in 2017-2018 that used an H1N1 seasonal flu virus and included a total of 31 volunteers aged 18 to 55; and the other in 2015 with 18 volunteers, mean age 21.7, that inoculated participants with human rhinovirus strain type 16.
In both studies, participants wore an E4 wristband (Empatica, Inc.) for a short time before nasal inoculation with the test virus, and then for several days afterwards. The E4 wristband measures heart rate, skin temperature, electrodermal activity, and movement.
The researchers defined symptoms as observable events, including fever, stuffy nose, runny nose, sneezing, coughing, shortness of breath, hoarseness, diarrhea and wheezy chest, and unobservable events, such as muscle soreness, fatigue, headache, ear pain, throat discomfort, chest pain, chills, malaise and itchy eyes.
For the data analysis, the researchers developed 25 binary, random forest classification models to predict infection versus non-infection, using features that came from the wristbands. Each model covered a different period of time post-inoculation, or used a different definition of infected and not infected.
The researchers found that their detection models, using only data from the wearable devices as input, were able to distinguish between infection and noninfection with accuracies of up to 92% for flu (90% precision, 90% sensitivity, 93%specificity, and 90% F1 score, 0.85 area under the curve) and 88% for rhinovirus (100% precision, 78%sensitivity, 100% specificity, 88% F1 score, and 0.96 AUC).
Their infection severity prediction model was able to distinguish between mild and moderate infection 24 hours prior to symptom onset with an accuracy of 90% for flu (88% precision, 88% sensitivity, 92% specificity, 88% score, and 0.88 AUC) and 89% for rhinovirus (100% precision, 75%sensitivity, 100% specificity, 86% F1 score, and 0.95 AUC).
The new study is “very interesting,” said Dr. Rutul Dalal, medical director, infectious diseases at UPMC in Pittsburgh, Pennsylvania. There have been studies like this measuring blood sugar in diabetic patients, Dr. Dalal said.
The problem with the current study is that there can be a lot of conditions that would cause temperatures to rise and heart rates to change, Dr. Dalal said. “So, I would probably take this with a pinch of salt,” he added.
Still, Dr. Dalal said, “it’s a step in the right direction. This is going to be the future of medicine since not everyone has access to get a test done as soon as possible. Hopefully there will be a randomized trial in the future that would include other conditions that could affect the results.”
And it’s important to know who has been infected ASAP to prevent the spread of infectious disease for community health protection, Dr. Dalal said.
SOURCE: https://bit.ly/3m6JWww JAMA Network Open, online September 29, 2021.
Source: Read Full Article