Study Testing Wearables Data for Covid-19 Clues

Originally published in Science & Enterprise

A biomedical engineering lab and analytics company are testing the feasibility of algorithms to find early signs of Covid-19 symptoms from smart watch users. The collaboration brings together the Cardiovascular Imaging Research Laboratory at Purdue University in West Lafayette, Indiana with PhysIQ, a digital health technology company in Chicago.

The Purdue University lab, led by biomedical engineering professor Craig Goergen, is conducting a study with PhysIQ to determine the feasibility of biometric data routinely collected by smart watches to provide the raw material for algorithms to screen individuals for Covid-19 symptoms. Studies conducted since the start of the pandemic support the idea that routine collection of data from wearable devices like fitness trackers or smart watches could provide an early screening of Covid-19 symptoms, without taking a separate diagnostic test.

Fitness trackers and smart watches routinely collect biometric data including heart rate, heart rate variability, and respiration rate, which under the right conditions, can provide early indicators of Covid-19 symptoms. Previous studies suggest viral infections increase resting heart and respiration rates and decrease heart rate variability before a patient develops a fever, says Goergen, but questions still remain whether a device worn on the wrist can reliably capture respiration rates, one of the key indicators.

“An increased heart rate or respiration rate means something different if it increased while you were resting as opposed to running,” says Goergen in a Purdue University statement, “but most smartwatches have difficulty distinguishing that. So it is really recovery and resting periods that we are focused on with this approach.”

PhysIQ develops health care analytics from continuous biosensor data in smart watches and wearable devices. The company’s analytics use deep machine-learning algorithms that capture an individual’s vital signs from wearables and smart watches, then provide a personalized report based on data from that person. PhysIQ was founded by Purdue alumnus Gary Conkright, and in January 2020 the company received a $500,000 award from the university’s foundry investment fund to advance it’s artificial intelligence technology.

In August, the Food and Drug Administration cleared PhysIQ’s remote monitoring algorithm to improve the accuracy of its assessments. FDA earlier cleared the company’s cloud-based analytics for heart rate, heart rate variability, atrial fibrillation detection, respiration rate, and personalized physiology change detection.

The joint study is recruiting some 100 participants on the Purdue campus to determine the feasibility of collecting relevant data with a smart watch. Participants are given a Samsung Galaxy smart watch with the PhysIQ app to collect data. Individuals will also be asked to wear a single-lead electrocardiogram sensor taped to their chests, to provide more conventional measures of heart rate and respiration. Data from participants are transmitted to PhysIQ’s accelerateIQ cloud-based analytical platform used to collect biosensor data for Covid-19 clinical trials.

A separate team led by Purdue electrical and computer engineering professor Fengqing Maggie Zhu will analyze the data to determine their adequacy for training algorithms for software to detect changes in vital signs indicating early Covid-19 symptoms.

The researchers are aiming for an early-warning screening system to alert users to the need for a Covid-19 diagnostic test. “There won’t be a point where a smart watch can tell you that you’re Covid-19 positive,” notes Goergen, “but it could potentially say, ‘Within the next couple of days, you might be getting sick and should go get tested.'”