FDA clears first AI device to spot hidden signs of COVID-19

Alongside the rapid development of vaccines, the FDA has cleared a number of COVID-19 breakthroughs for emergency use as part of the ongoing fight against the devastating virus. So far, we've seen the agency approve medical advances including lab-made monoclonal antibodies for moderate infections that risk turning more severe, a rapid test that uses CRISPR gene-editing tech and Fitbit's Flow ventilator. The latest tool to gain clearance is the first AI-based screening device designed to pinpoint lurking signs of COVID-19 in asymptomatic people.

Dubbed the Tiger Tech COVID Plus Monitor, the apparatus is an armband that uses light sensors and a small computer processor to check for biomarkers of the virus, such as hypercoagulation — a common COVID-19 abnormality that causes the blood to clot more easily. Once strapped to a person's arm, the monitor's onboard sensors start collecting pulse signals from blood flow over a period of three to five minutes. The processor then extracts the key info from the measurements and feeds them through the machine learning model. Final results, including whether the test shows positive biomarkers (like hypercoagulation) or is inconclusive, are represented by different colored lights.

In its announcement, the FDA highlights that the armband is not meant to diagnose COVID-19 and therefore should not act as a substitute for a regular test. Nor is it designed for use on symptomatic people. Instead, it's intended as more of a back-up or precaution to be deployed alongside a temperature check if the reading doesn't indicate a fever in potentially asymptomatic individuals. Together, the two screening tools could be used to curb the spread of the virus in a range of public settings including healthcare facilities, schools, workplaces, theme parks, stadiums and airports, the agency said.

The monitor's clearance follows clinical studies in hospitals and schools that showed similar results. In the hospital setting, the device correctly spotted COVID-19 biomarkers at a rate of 98.6 percent, while also accurately identifying those that did not possess the telltale signs at a rate of 94.5 percent. The school trial essentially validated the results, according to the FDA.