AI Finds Gene Expression Data to Predict COVID-19 Patient Outcomes

An artificial intelligence algorithm can parse gene expression data to predict patient outcomes by identifying signature genes that reveal the severity of viral infection immune response including COVID-19, according to a recent study published in eBiomedicine.

University of California San Diego School of Medicine researchers analyzed over 45,000 transcriptomic datasets from viral pandemics among humans, mice, and rats. They identified a 166-gene signature that sheds light on how the immune system reacts to viral infections. In addition, a signature of 20 genes can predict the severity of a patient’s condition.

"When the COVID-19 pandemic began, I wanted to use my computer science background to find something that all viral pandemics have in common—some universal truth we could use as a guide as we try to make sense of a novel virus," said Debashis Sahoo, PhD, one of the study’s authors and an assistant professor of pediatrics at UC San Diego School of Medicine and of computer science and engineering at Jacobs School of Engineering, in a press release.

"This coronavirus may be new to us, but there are only so many ways our bodies can respond to an infection."

The gene signatures “provide a quantitative and qualitative framework for titrating the immune response in viral pandemics and may serve as a powerful unbiased tool to rapidly assess disease severity and vet candidate drugs,” the study stated.

Researchers looked for patterns in viral respiratory infections including swine flu, MERS, and SARS, and validated the algorithm using lung tissues from deceased patients. The 166-gene signature also performed well with identifying and classifying bacterial and fungal infections, which shows promise for the AI model’s effectiveness outside of coronaviruses.

Researchers defined the severity of a disease by ranking the 166 genes. Intubation and medical ventilation were defined as elements of severe disease. With each new dataset that became available, researchers tested the model and found the same gene expression patterns in each case.

"These viral pandemic-associated signatures tell us how a person's immune system responds to a viral infection and how severe it might get, and that gives us a map for this and future pandemics," explained Pradipta Ghosh, MD, one of the study’s authors and a professor of cellular and molecular medicine at UC San Diego School of Medicine and Moores Cancer Center, in the press release.

The study’s primary finding is that viral pandemics, regardless of their differences, all share a fundamental host immune response. The 166-gene signature revealed a phenomenon called a cytokine storm, in which the body releases too many cytokines, causing the immune system to attack healthy tissue.

Researchers were also able to define the source of the cytokine storm and revealed that the storm can lead to lung airway cells being damaged, preventing the immune system from killing the cells infected by the virus.

"We could see and show the world that the alveolar cells in our lungs that are normally designed to allow gas exchange and oxygenation of our blood, are one of the major sources of the cytokine storm, and hence, serve as the eye of the cytokine storm," explained Soumita Das, PhD, co-author and associate professor of pathology at UC San Diego School of Medicine.

The study acknowledged that as new COVID-19 datasets emerge, the artificial intelligence model will be even more effective and accurate.

"It is not a matter of if, but when the next pandemic will emerge," continued Ghosh in the press release. "We are building tools that are relevant not just for today's pandemic, but for the next one around the corner."