AI Pioneers a Public Health Revolution by Predicting Viral Evolution
Paul Grieselhuber
Founder, director
The prospect of being able to anticipate viral transformations stands as a beacon of promise for public health, suggesting a future where vaccine development can operate on a proactive basis, and antiviral strategies are crafted with foresight. At the forefront of this visionary landscape, artificial intelligence (AI) shines brightly, promising to decode the complexities of viral behavior. Around the globe, research groups are wielding AI as a tool to unravel the genetic mysteries of viruses such as SARS-CoV-2 and influenza.
Central to our understanding is the utilization of AI to sift through genetic codes, allowing us to predict potential mutations and ensuing variants. In doing so, we edge closer to a reality where we are not perpetually lagging behind viruses' rapid adaptation. AI innovations such as AlphaFold, ESM-2, and CoVFit have been instrumental in these pursuits. They draw upon expansive datasets to forecast viral shifts: consider the nearly 17 million SARS-CoV-2 sequences available today. Such resources serve as the foundation from which these models foresee the viral landscape's evolution.
For instance, Harvard Medical School's EVEscape model stands as testament to the power of AI, with its engineered SARS-CoV-2 spike protein variants that can escape currently known antibodies, thus enabling preemptive testing of vaccine efficacy. Similarly, the CoVFit model from the University of Tokyo heralded the model's prowess by accurately predicting the rise of SARS-CoV-2 variants, like the JN.1 strain which dominated the global stage in 2024.
The journey, however, is dotted with continuous hurdles. Unexpected evolutionary surges, akin to the emergence of the Omicron variant, present formidable challenges. While AI has become adept at tracking incremental viral adjustments, the unpredictable leaps in viral evolution call for innovative methodologies to chart these extensive possibilities. Bridging sequencing data with experimental insights promises to be a fertile ground for enhancing these AI models, equipping researchers with a comprehensive view of viral evolution.
The mission carries immense weight: as evolutionary virologists like Shusuke Kawakubo delve into the immune-evasion capabilities of influenza's haemagglutinin protein, the end goal looms clear—to outpace viral evolution and maintain vaccine potency. AI's integration into this domain signifies a monumental shift towards defending global health, and while much work lies ahead, we are at the cusp of a transformative era.
Further Reading
References:
- Smriti Mallapaty (2025). "What will viruses do next? AI is helping scientists predict their evolution." Nature. Available online: doi: 10.1038/d41586-024-04195-3. Accessed 16 January 2025.
