MONDAY, March 30, 2020 (HealthDay News) -- Data from patients with coronavirus disease 2019 (COVID-19) can be used to develop a model that can predict who will develop acute respiratory distress syndrome (ARDS), according to a study published online March 30 in Computers, Materials & Continua.
Xiangao Jiang, M.D., from Wenzhou Central Hospital in China, and colleagues presented a first step toward building an artificial intelligence framework, with predictive analytics capabilities applied to real patient data in order to provide support for rapid clinical decision making. Learning from historical data from 53 hospitalized patients with COVID-19 from two Chinese hospitals, the models aim to help predict who will develop ARDS.
The researchers found that experimental results identified features most predictive of ARDS in COVID-19 initial presentation, which would not have been obvious to clinicians. The features on presentation that were most predictive were a mild increase in elevated alanine aminotransferase, myalgias, and an increase in hemoglobin, in that order. For predicting severe cases, models that learned from historical patient data achieved 70 to 80 percent accuracy.
"Our goal was to design and deploy a decision-support tool using artificial intelligence capabilities -- mostly predictive analytics -- to flag future clinical coronavirus severity," a coauthor said in a statement. "We hope that the tool, when fully developed, will be useful to physicians as they assess which moderately ill patients really need beds, and who can safely go home, with hospital resources stretched thin."