Researchers at the Massachusetts Institute of Technology (MIT) have engineered an Artificial Intelligence (AI) model that can anticipate early on which SARS-CoV-2 variants might trigger fresh waves of infections.
Under the leadership of Retsef Levi from MIT’s Sloan School of Management, the team delved into the elements influencing viral transmission. They analyzed a vast dataset comprising 9 million SARS-CoV-2 genetic sequences sourced from 30 nations by the Global Initiative on Sharing Avian Influenza Data (GISAID). This data was combined with information on vaccination rates, infection frequencies, and other relevant metrics.
Their research outcomes have been documented in the journal PNAS Nexus.
From the data analysis, distinct patterns surfaced. These insights were harnessed to construct a risk assessment model powered by Machine Learning.
Remarkably, this model could identify 72.8% of the variants in any given country that might lead to at least 1,000 infections per million individuals within three months, merely a week post-detection. This accuracy rose to 80.1% after two weeks.
Key indicators suggesting the infectious potential of a variant include its initial infection trajectory, spike mutations, and the genetic divergence from the prevalent variant during the studied period.
“This work provides an analytical framework that leverages multiple data sources, including genetic sequence data and epidemiological data via machine-learning models to provide improved early signals on the spread risk of new SARS-CoV-2 variants,” said the researchers in the study.
While underscoring the need for further investigation, the researchers highlighted the potential applicability of their modeling strategy to other respiratory viruses, such as influenza, avian flu, or other strains of coronaviruses.
