Can artificial intelligence predict how sick it will be from COVID-19?Scientists think so | National

San Diego — A team of San Diego scientists are using artificial intelligence to understand why the symptoms of COVID-19 vary dramatically from person to person. This information may help in the ongoing fight against the coronavirus and future pandemics.

Researchers looked at publicly available data to see how other viruses change which genes a cell turns on or off. Using that information, they discovered a set of genes that were activated across a wide range of infectious diseases, including the new coronavirus. These genes predicted whether someone had mild or severe cases of COVID-19 and whether they were likely to be hospitalized for a long time.

A team led by the University of California, San Diego, joined with researchers from Scripps Research and the La Jolla Institute for Immunology, and announced the findings on June 11. The authors of the study say their approach helps determine if new treatments and vaccines are working.

“When the whole world faced this pandemic, it took people months to understand the new virus,” said Dr. Pradipta Gauche, a cell biologist and author of the study at the University of California, San Diego. I did. “I think we need more of this computational framework to get into this panic situation.”

The project began in March 2020, with Ghosh working with computer scientist Debashis Sahoo at the University of California, San Diego, where the new coronavirus causes little or no symptoms in some people and disrupts others. I got a better understanding of why it came.

There was only one problem. The new coronavirus was, well, new. In other words, there wasn’t much data to learn.

So Sahoo and Ghosh took a different approach. They accessed a public database and downloaded 45,000 samples from various viral infections such as Ebola, Zika, influenza, HIV and hepatitis C virus.

Their hope was to find a common response pattern to these viruses. That’s exactly what they saw. 166 genes that were consistently cranked up during infection. In the list, 20 genes generally distinguished patients with mild symptoms from those with severe illness.

Coronavirus was no exception. Sahoo and Ghosh stated that they identified this common viral response pattern before testing with samples from COVID-19 patients and infected cells, but the results were surprisingly well maintained.

“It seemed to work with all the datasets we used,” Sahoo said. “I couldn’t believe it.”

Their findings indicate that respiratory failure in COVID-19 patients is the result of overwhelming inflammation that damages the airways and reduces the effectiveness of immune cells over time.

Stanford University’s Pervesh Katri is not surprised. In his lab, he uses computer algorithms and statistics on a daily basis to find patterns in large sets of immune response data. In 2015, Khatri’s group discovered that the respiratory virus provokes a common reaction. And in April, they reported that this shared response was also applied to a variety of other viruses, including the new coronavirus.

Katri says it makes sense because researchers have long known that there are specific genes that turn on the immune system in response to virtually any viral infection.

“Overall, the idea is pretty solid,” said Katri, a recent study led by the University of California, San Diego. “All genes are () normal suspects.”

Sahoo and Ghosh continue to test new coronavirus data as it becomes available. They are particularly interested in COVID-19 long haul carriers. Gauche says he has already seen people with prolonged coronavirus symptoms have a clear pattern of gene activation compared to those who have fully recovered. Think of it as an indelible smoldering fire.

The ultimate hope of researchers is not only to predict and understand serious illnesses, but to stop them. For example, if a blood sample suggests that the current treatment is likely to get sick, doctors can give the patient another treatment, they say. Gauche said the genetic patterns they see could help identify promising new treatments and vaccines for future pandemics based on treatments that prevent reactions associated with serious illness. I add that there is.

“In the unknown, unknown territory, this provides a guardrail for looking around, understanding (viruses), finding solutions, building better models, and finally finding cures.”