News | Feb. 10, 2021

Machine Learning: DTRA Helps to Equip for the Unknown and Uncertain

FORT BELVOIR, Va. - In the midst of the COVID-19 pandemic, the Department of Defense (DoD) offers a new tool in the fight against the disease -- the ability to warn of an infection up to 48 hours before overt symptoms appear. The Defense Threat Reduction Agency is developing an early infection-warning system that is grounded on the clinical fact that when a person is exposed to a pathogen, that person undergoes certain measurable changes in the body before symptoms appear. This infection-warning capability is called the Rapid Analysis of Threat Exposure (RATE).

“RATE is a machine-learning algorithm that can explore a large set of data for patterns in how the human body changes when it acquires an infection,” said Dr. Chris Kiley, a DTRA scientist. “The algorithm non-invasively collects biomarkers to identify signs of an infection regardless of the type of pathogen causing the infection.”

The algorithm can analyze data collected by wearable technologies, such as smart watches, to identify individuals who are about to develop overt symptoms due to exposure to pathogens such as SARS-CoV-2. Armed with this knowledge, these individuals can seek medical help and self-quarantine to prevent disease transmission. If military personnel were outfitted with wearable technology, then RATE’s threat-agnostic, infection-warning capability could help DoD better prepare for surprises due to biological threats.

“While RATE isn’t preventing disease, it informs of an illness about to overwhelm the body, in turn promoting behavior that can assure a quicker return to wellness,” said DTRA’s Dr. John Hannan. “A 48-hour warning of an infection, regardless of the type of pathogen causing it, enables military personnel to initiate self-care, seek medical help, and self-quarantine.”

Although RATE was developed using high quality, medical grade data from patients who developed infections during hospital stays, it is now being assessed for its ability to indicate early signs of infection in active and otherwise healthy military personnel. The assessment is making use of wearable devices with SARS-CoV-2 as the infection-causing pathogen of interest, due to the assessment coinciding with the COVID-19 pandemic.
Through the ongoing RATE COVID-19 Demonstration, DTRA is evaluating the algorithm’s ability to utilize data gathered in nonclinical settings from wearable technologies to warn of an infection while maintaining confidentiality of personal identity and physical location.

“RATE is the first machine-learning, infection-warning algorithm of its kind, offering DoD a novel opportunity to get ahead of illness among its military personnel,” said Hannan. “With better data, action is possible and as the current pandemic continues to proliferate, RATE provides a distinct advantage for our warfighters.”