What Does Military AI Need to Detect COVID? Data From Wearables

 In Intelligence, Industry, Acquisition, & Innovation, COVID-19, Defense

WASHINGTON — The U.S. Defense Department is expand­ing a research study that detects poten­tial COVID-19 infec­tions through wear­able devices to “sev­er­al thou­sand par­tic­i­pants,” offi­cials announced Tuesday.

The project, from the Defense Innovation Unit, Defense Threat Reduction Agency (DTRA) and Philips Healthcare uses arti­fi­cial intel­li­gence to ana­lyze bio­met­ric data to detect poten­tial COVID-19 infec­tions in asymp­to­matic and pre-symp­to­matic car­ri­ers and slow the spread of the virus.

The early-warn­ing system pro­to­type is part of the DoD’s Rapid Analysis of Threat Exposure (RATE) tech­nol­o­gy and has been used for COVID-19 detec­tion for a dozen dif­fer­ent cohorts of mil­i­tary and non-mil­i­tary per­son­nel in clinic trials, accord­ing to a press release. The system can detect early, “sub-clin­i­cal” symp­toms up to 48 hours before overt symp­toms appear, Lt. Col. Jeff “Mach” Schneider, a DIU pro­gram man­ag­er told C4ISRNET.

“The imme­di­ate pri­or­i­ty was to work with cohorts that were either in some kind of front line capac­i­ty,” such as health care work­ers, “or that were more severe­ly impact­ed from a readi­ness stand­point,” he said.

The tech­nol­o­gy is in use by about 700 work­ers in the U.S. Navy and Office of the Secretary of Defense, Schneider said. He added that the study will expand to about 5,000 new users across dif­fer­ent ser­vices, com­mands and the United States Military Academy.

“As we con­tin­ue to col­lect data and refine the algo­rithm, pri­or­i­ty will con­tin­ue to be pro­vid­ed to those first respon­ders and those that have had to adopt new oper­a­tional tempo to sup­port their duty oblig­a­tions,” Schneider said.

The RATE pro­gram uses machine learn­ing and arti­fi­cial intel­li­gence to ana­lyze a person’s bio­met­ric data and detect small changes in a person before they expe­ri­ence any symp­toms. The system can then alert a user they may be infect­ed so they can quar­an­tine or change their behav­ior to pre­vent fur­ther spread­ing the dis­ease.

“RATE would allow us to non-inva­sive­ly mon­i­tor a ser­vice member’s health and pro­vide early alerts to poten­tial infec­tion that will help us to ensure troop readi­ness, better sup­port their health and pro­tect against the threat of fur­ther spread of the dis­ease,” said Edward Argenta, sci­ence and tech­nol­o­gy man­ag­er at DTRA, said in a state­ment. “Unlike other more narrow approach­es, this solu­tion is designed to rec­og­nize a wide vari­ety of infec­tions and can help iden­ti­fy future novel threats.”

#nl-modal-dialog{ – header-text-color:#D2232A}

The RATE program’s COVID-19 detec­tion appli­ca­tion deployed to mil­i­tary units in mid-June, while Dr. Joseph Frassica, Philips’ head of research in the Americas, added that the algo­rithm is being fur­ther refined to better project COVID-19 infec­tions. The RATE-COVID app is designed to work with all wear­ables and the team is look­ing to deploy to bring-your-own devices “in the future,” the release said.

RATE orig­i­nal­ly start­ed years ago to slow the spread of gen­er­al infec­tions such as the flu, Frassica told C4ISRNET in an inter­view. In March, the part­ner­ship piv­ot­ed to the coro­n­avirus pan­dem­ic, he said.

“With Covid it became even more crit­i­cal to find those people who are pre-symp­to­matic to get them out of the pop­u­la­tion before they have a high enough viral load to start to spread the dis­ease,” Frassica said.

The RATE pro­gram uses large-scale machine learn­ing to ana­lyze data across 165 bio­mark­ers from a data set owned by Philips that con­tains 41,000 cases of hos­pi­tal-acquired infec­tions. Biomarker data is then process and ana­lyzed in the cloud, gen­er­ates an hourly RATE score that ser­vice mem­bers can then access through a secure web­site. it looks at sev­er­al bio­met­ric mark­ers such as tem­per­a­ture and blood pres­sure. The press release added that the risk score also works for “mul­ti­ple gen­er­al types of infec­tion,” with Philips sci­en­tists also work­ing to gen­er­ate a tool that can detect the “next new unknown infec­tious agent.”

“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, DIU human sys­tems port­fo­lio direc­tor, in a state­ment.

Schneider said that that project cost about $7 mil­lion in total, includ­ing fund­ing from the Defense Advanced Research Projects Agency and Coronavirus Aid, Relief, and Economic Security Act. Texas A&M’s Engineering Experiment Station also worked on the project.

C4ISRNET source|articles

Recommended Posts

Start typing and press Enter to search