MIT researchers created an AI model to forecast new SARS-CoV-2 infections. The model identifies virus propagation determinants using 9 million genomic sequences from 30 nations.
Researchers at Massachusetts Institute of Technology (MIT) have created a new AI model. This cutting-edge approach can predict which SARS-CoV-2 mutations will cause new illnesses. The idea advances pandemic management by detecting and responding to surges early.
Models for viral transmission dynamics do not anticipate variant-specific dissemination.
The research team, led by MIT Sloan School of Management’s Retsef Levi, analyzed 9 million SARS-CoV-2 genetic sequences from the Global Initiative on Sharing Avian Influenza Data. The study examined vaccination rates, infection rates, and other characteristics from 30 nations to determine the virus’s spread. The findings appear in PNAS Nexus.
This investigation produced patterns for a Machine Learning-enabled risk assessment model. After one week of observation, the algorithm can detect 72.8% of variants in each country that will produce at least 1,000 instances per million people in the next three months. Two weeks of observation improves prediction to 80.1%. The variation’s spike mutations, early infection trajectory, and how dissimilar a new variant’s mutations are from those of the most dominant variant during the observation period are strong predictors of infectiousness.
The study authors noted, “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.”
They suggested applying the modelling approach to other respiratory viruses such influenza, avian flu, and coronaviruses, but additional research was needed.
Conclusion
MIT researchers created an AI model to forecast SARS-CoV-2 mutations that cause new illnesses. Based on 9 million genetic sequences from 30 nations, the program can detect 72.8% of variations that will cause 1,000 instances per million individuals in three months. Two weeks of observation improves the model’s prediction to 81.1%. The researchers advise applying this method to additional respiratory viruses like influenza, avian flu, and coronaviruses to better early detection and response to surges.
