When individuals who appear to be in good health receive a notification from a wearable device indicating that they might have a respiratory virus—resulting from minute variations in their specific heart rate, sleep, and activity levels—what actions do they take? A recent study from Scripps Research, conducted during the peak of the COVID-19 pandemic, found that only about 25% of people acted on such an alert by conducting an at-home viral test.
When individuals who appear to be in good health receive an alert from a wearable device indicating they might have a respiratory virus—based on subtle variations in their specific heart rate, sleep, and activity patterns—what actions do they take? A recent study by scientists at Scripps Research, conducted during the peak of the COVID-19 pandemic, revealed that merely a quarter of participants follow up such an alert with an at-home viral test.
This finding is one conclusion of the new research published in The Lancet Digital Health on July 24, 2024. The study explored the effectiveness of wearable sensors and the related alerts in guiding people’s actions. The results highlighted the crucial role of timing and personalized communication of alerts, which are essential for developing and implementing new wearable technologies that monitor various health metrics, from infections and blood sugar levels to menstrual cycles and pregnancies.
“The most exciting aspect of this study is that we demonstrated the potential to provide a personalized infection alert using data passively collected by sensors,” explained Giorgio Quer, PhD, the lead author and director of artificial intelligence at Scripps Research Translational Institute. “We also identified key challenges in effectively delivering these alerts to participants, emphasizing the need to communicate these changes in a way that encourages individuals to take appropriate actions.”
In an earlier study, Quer and his team learned that data from fitness and health trackers could accurately predict the likelihood of a COVID-19 infection with about 80% accuracy. Indicators such as increased sleep, decreased physical activity, and elevated resting heart rates were linked to the presence of respiratory infections.
The current study investigated whether notifying individuals about these physiological changes would motivate them to perform at-home tests for COVID-19 and respiratory syncytial virus (RSV), even in the absence of symptoms. Between September and December 2021, researchers enrolled 450 adults across the United States. Participants were divided into three groups: one that received alerts prompting testing based on sensor data changes or reported symptoms (both the sensor and tests were provided), one that received alerts solely based on reported symptoms, and the final group that did not receive any alerts or tests.
“Notifying individuals about early physiological signals suggesting a viral infection can lead to both personal and public health advantages,” stated Steven Steinhubl, MD, the study’s senior author and an adjunct at Scripps Research as well as a professor of Biomedical Engineering at Purdue University. “This early alert can offer individuals time to isolate, adjust their plans, and help prevent the spread of the virus.”
Throughout the study, 118 participants (39%) were prompted to conduct a self-test at least once, and 62 of these individuals (52%) successfully completed the test and recorded their results. More frequent self-testing was tied to symptom alerts, as participants were more likely to test when they were already experiencing symptoms; only 23% tested when alerted by sensor data changes, compared to 56% when alerted by their symptoms.
Researchers suspect one reason for this trend is likely due to the timing of alerts based on sensor data, which were communicated at a set time in the morning when participants may not have been readily available to test. In contrast, symptom-based alerts were sent when symptoms were reported, aligning with times when individuals were considering the possibility of a virus and likely had a few spare moments.
“This emphasizes the importance of considering the timing of health alerts for future wearable monitoring studies,” emphasized Quer. “The moment at which you notify individuals of health changes is crucial for influencing their behavior.”
Among other insights gained from this digital health feasibility trial: researchers successfully recruited a diverse participant pool without requiring in-person visits to medical facilities. However, accessing infection data from the provided electronic medical records proved more challenging than anticipated.
Quer notes that these insights are crucial for shaping future research utilizing wearable health monitoring devices. His team is currently investigating whether similar sensors can notify pregnant women about heightened risks for health issues such as gestational diabetes, pre-eclampsia, or premature deliveries.
“We must focus not only on making accurate predictions but also on returning information to patients in a way that is meaningful,” he added.
