VOA科学技术2023--AI Tool Aims to Predict Virus Variants before They Appear（在线收听）

AI Tool Aims to Predict Virus Variants before They Appear

Scientists have created an artificial intelligence (AI) tool that aims to predict different virus versions, called variants, before they appear.

Creators of the tool say it works by estimating the likelihood that a variant will permit it to escape the body's own immunity defenses.

The tool is called EVEscape. It works by processing collected data on the development, or evolution, of a virus over time. The system also considers detailed biological and structural information related to the virus.

The method is a form of "generative AI," a term describing technology tools designed to operate at human levels. EVEscape was developed by researchers from Britain's University of Oxford and Harvard Medical School in Boston, Massachusetts. The team's findings recently appeared in a study in the publication Nature.

Many viruses repeatedly mutate in an effort to stay alive and strong. This requires viruses to infect living organisms. Once inside the body, viruses continue to reproduce and spread.

The U.S. Centers for Disease Control and Prevention (CDC) explains that repeated mutating creates new variants. Some of these mutations permit the virus "to spread more easily or make it resistant to treatments or vaccines," the CDC said.

The researchers built on previous work involving a tool they created called EVE. This tool was designed to study gene mutations that can cause human disease. In a past study, the team said it successfully used EVE to identify mutations linked to conditions like cancer and heart disease.

During the COVID-19 pandemic, the scientists said they recognized a good chance to "rebuild" EVE into a new tool "for the purpose of predicting viral variants."

To do this, the team explained it "turned the clock back to January 2020, just before the COVID-19 pandemic started." The researchers asked the EVEscape tool to predict what mutations would develop from the SARS-CoV-2 virus that causes COVID-19.

The researchers reported the tool correctly predicted which SARS-CoV-2 mutations would appear and which would become most common, or dominant.

The team also said the tool predicted which antibody-based treatments "would lose their efficacy as the pandemic progressed and the virus developed mutations to escape these treatments."

The study noted EVEscape can make predictions more quickly and effectively than lab-based methods. This is because the tool does not need "to wait for relevant antibodies to arise in the population and become available for testing."

The team said the system was able to examine tens of thousands of new virus variants appearing each week and identified the ones most likely to become problematic.

Sarah Gurev is a student in the Electrical Engineering and Computer Science program at the Massachusetts Institute of Technoology (MIT). She was a leader of the research. Gurev said that by quickly identifying the threat level of new variants, "we can help inform earlier public health decisions."

The team said it continuously uses the system to examine real-time data about the development of new variants of SARS-CoV-2. It then shares this information with the public and international agencies such as the World Health Organization (WHO).

The researchers said their experiments also showed the EVEscape tool could successfully predict mutating behaviors for other common viruses, including HIV and the flu.

Debora Marks is a professor of systems biology at Harvard Medical School's Blavatnik Institute. She was a lead writer of the study. She said the researchers would continue to improve the tool's ability to predict new variants for a number of viruses. "Because if we can, that's going to be extremely important for designing vaccines and therapies."

Words in This Story

immunity – n. bodily power to resist an infectious disease

mutate – v. to cause a gene to change and create an unusual characteristic in a plant or animal

efficacy – n. the ability to produce a desired result

relevant – adj. related or useful to what is happening or being talked about

therapy – n. a treatment method for disease