DIU’s latest health project: AI-powered early coronavirus detection

The military’s Silicon Valley innovation office is working on an artificial intelligence-enabled early warning system for detecting COVID-19 infections in military personnel.

The project at the Defense Innovation Unit pulls biometric data from wearable devices, like smartwatches, that can detect slight physiological changes in a person who has contracted the disease but is not symptomatic. With early detection, DIU hopes it can help stop the spread of infection within the military.

Called Rapid Analysis of Threat Exposure (RATE), the project is being developed in partnership with the Defense Threat Reduction Agency (DTRA) and Philips North America. The initial study of the system started in June, with an expanded group of participants joining the program in the coming weeks.

“By combining commercial technology, a rich data source and simple-to-use wearables, we are effectively providing a check-engine light on the military service member and getting that alert before they’re broken down with a disease. In military speak, we’re targeting left-of-cough awareness,” said Christian Whitchurch, director of DIU’s Human Systems Portfolio.

Health-related projects pop up occasionally at DIU, which is better known for working in areas like drones or cyberthreat detection. Earlier this month the agency announced a cancer detection project with Google.

One RATE for all

The dataset for the new project’s machine learning came from 165 biomarkers in 41,000 COVID-10 cases. As an “early warning system,” RATE reads from the wearable to give an indication of infection before a person develops symptoms like coughing, fever and a loss in taste or smell. The goal is to create the algorithms and data structure to work with any wearable device that can pull biometric data.

So far, DIU says the tests have already yielded positive results and “multiple” cases of COVID-19 have been caught early.

“While it is difficult to say how many [infections] were stopped” Lt. Col. Jeffrey “Mach” Schneider, a DIU program manager said in a statement, “this is an extremely useful tool for commanders to quickly isolate a COVID outbreak from rapidly spreading throughout their unit,”

For now, the system is only being used on a few thousand military personnel. A volunteer’s data is de-identified and is not shared with Philips. Users of RATE system have a secure web portal where they can check their “score,” or potential signs of an infection to try and stop asymptomatic or presymptomatic spread of the disease.

“The technology could further be applied in a civilian capacity by helping to monitor hospital patients for infection prior to clinical symptoms,” the release states.

The system could also be retrained to have early detection for other diseases, according to Philips.

“The RATE science shows that physiological response to infection has similarities across
different types of infectious agents, and we anticipate that this will also apply to RATE-COVID,
giving us a useful early warning solution,” said Dr. Joe Frassica, chief medical officer and head of Philips Research North America.