VA Pilots New Tool to Better Understand COVID Mortality Rates

The Department of Veterans Affairs (VA) is piloting a new artificial intelligence tool that can predict a COVID-19 patient’s prognosis and recommend next steps. Especially during the COVID-19 pandemic, hospitals have seen an influx of patients, and doctors and nurses often have to make quick, life-saving decisions in order to help their patients.

Currently, even after a positive COVID diagnosis, doctors are not able to predict whether or not the patient will need hospitalization or how rapidly the virus will spread. Gil Alterovitz, director of AI at the VA said, “The problem is, given a positive detection of COVID viral test—if someone has a positive test—what is the prognosis? Will the patient need hospitalization? Are they at risk of death, etc.? What will be their needs? The approach was to develop an applied AI model for that using both clinical and nonclinical information.”

This tool began development in spring. Data from thousands of COVID patients in the VA system has been collected, including information such as blood pressure, fever, and other symptoms. This information was then analyzed to see how the symptoms affected mortality rates.

Veterans Health Administration (VHA) Senior Innovation Fellow Amanda Purnell explained, “AI will be used to mine the data to look for predictive factors, including analyzing predictive value of each factor to then develop operation tools to guide clinical decisions within the VHA."

For example, if a patient has a preexisting condition such as heart problems or diabetes, doctors may believe the person has a higher risk of mortality. However, upon analyzing other factors that the AI tool shows, it can be concluded that the person’s mortality rate is not as high as previously suspected.

Mr. Alterovitz continued, “It’s being used already operationally. And we’re going to be now working with other medical sites to see how this can be used across the country in different settings, as well.”