AI could predict and warn before next pandemic breaks out
Scientists have developed an Early Warning Anomaly Detection (EWAD) system that accurately predicts new variants of concern (VOCs) in future pandemics. Using machine learning, the AI analyzes genetic sequences of SARS-CoV-2 variants, infection rates, and mortality rates to spot genetic shifts weeks before the WHO officially designates VOCs. The researchers discovered "rules" of virus evolution that could be crucial in combating emerging pandemics.
Unraveling virus evolution with machine learning
The EWAD system was developed using a machine learning method called Gaussian process-based spatial covariance. It predicts new data by analyzing existing data points and their relationships. By testing the model on real events and finding close matches between real and predicted data, scientists proved EWAD's effectiveness in predicting how measures like vaccines and mask-wearing impact virus evolution.
AI system could be game-changer in pandemic preparedness
The EWAD system could help scientists understand virus biology, leading to improved treatments and public health measures. The "rules" of virus evolution identified by the AI algorithm could be vital in fighting future pandemics. Mathemologist Ben Calverley from Scripps Research believes the system and its underlying technical methods could have numerous potential applications in the future.