- To forecast which versions of the SARS-CoV-2 virus, a novel AI model has been created.
- In each country, the model can identify approximately 73% of variants.
- The innovative modeling technique may be used to forecast the future course of various infectious diseases
To forecast which versions of the SARS-CoV-2 virus, which causes COVID, would probably lead to subsequent infection waves, a novel Artificial Intelligence (AI) model has been created.
In each country, the model can identify approximately 73% of variants that will result in at least 1,000 instances per 10 lakh people within three months after a one-week observation period and more than 80% after two weeks.
AI modeling technique
Scientists from the Hebrew University Hadassah Medical School and the Massachusetts Institute of Technology examined nine SARS-CoV-2 million genomic sequences from thirty different nations that were gathered by the Global Initiative on Sharing Avian Influenza Data (GISAID).
Utilizing the patterns that emerged from the investigation, the team developed a machine-learning-based risk assessment model, which is an artificial intelligence program capable of learning from historical data and forecasting future events.
The early course of the infections a variant generated, its spike mutations, and the degree to which its mutations differed from those of the most dominant variant during the observation period were the best indicators of a variant’s infectiousness.
These findings corroborate the theory that infectious novel variations are those that have enough mutations to allow for the targeting of previously immune-naive population subgroups or the possibility of reinfections.
The innovative modeling technique may be used to forecast the future course of various infectious diseases and extend to other respiratory viruses like influenza, avian flu viruses, or other coronaviruses.
Future studies could examine how knowledge of a variant’s infectiousness and dissemination derived from genetic and biological research can be turned into predictive criteria that are assessed using the data at hand.
