Scientists from the University of Oxford, together with colleagues from academia, industry and policy organisations from Africa, the Americas, Asia, Australia and Europe, investigated the potential of AI as a tool to better prepare the world in the event of the next pandemic. In particular, the research focused on safety, accountability and ethics in the deployment and use of AI in infectious disease research.
In their study, the scientists and academics call for collaboration and transparency, both in terms of datasets and AI models. Currently, medical AI applications focus mainly on the care of individual patients, improving, for example, clinical diagnostics, precision medicine or supporting clinical treatment decisions. The study, published in Nature, describes how advances in AI can accelerate breakthroughs in infectious disease research and outbreak response.
Ever-improving AI methodologies
The study finds that recent developments in AI methodologies are performing increasingly well, even with limited data. Something that is still a major bottleneck to date. Better performance based on noise and limited data opens up new areas for AI tools to improve health, both in rich ‘Western’ countries and in developing countries.
‘In the next five years, AI has the potential to transform pandemic preparedness. Using terabytes of routinely collected climatic and socio-economic data, we can better anticipate where outbreaks will start and predict their course. It could also help predict the impact of disease outbreaks on individual patients by studying the interactions between the immune system and emerging pathogens. Taken together, and if integrated into countries‘ pandemic response systems, these advances could save lives and ensure that the world is better prepared for future pandemic threats,’ said Professor Moritz Kraemer of Oxford University's Pandemic Sciences Institute.
Better preparing for a pandemic with AI
The study identifies several options for how AI can help to better prepare for a possible next (global) pandemic, including:
- Promising progress in improving current models of disease spread, with the aim of making modelling more robust, accurate and realistic.
- Progress in identifying areas with high infection potential so that limited healthcare resources can be used as efficiently as possible.
- Potential to improve genetic data in disease surveillance, ultimately accelerating vaccine development and identification of new variants.
- Potential to help determine the properties of new pathogens, predict their traits and identify whether cross-species jumps are likely.
- Predict which new variants of already circulating pathogens, such as SARS-CoV-2 and influenza viruses, may emerge and which treatments and vaccines can best mitigate their impact.
- Potential AI-assisted integration of population-level data with data from individual-level sources - including wearable technologies such as heart rate and step counts - to better detect and monitor outbreaks.
- AI can create a new interface between highly technical science and healthcare professionals with limited training, improving capacity in environments that need these tools the most.
AI alone is not enough
However, not all areas of pandemic preparedness and response will be equally impacted by advances in AI. For example, while protein language models are promising for faster understanding of how viral mutations can affect disease spread and severity, advances in basic models may yield only modest improvements over existing approaches for modelling the rate at which a pathogen spreads.
The scientists are also cautious about suggesting that AI alone has sufficient potential to address the challenges associated with infectious diseases and containing a possible pandemic. They argue that human feedback will continue to play an important factor and for addressing and overcoming existing limitations of AI models. In the study, the scientists are also particularly concerned about the quality and representativeness of training data, the limited accessibility of AI models to the wider community and the potential risks associated with using black-box models to make decisions.
‘While AI has remarkable transformative potential for pandemic control, it depends on extensive global collaboration and on comprehensive, continuous surveillance data,’ says Professor Eric Topol, MD, founder and director of the Scripps Research Translational Institute.
That AI is a valuable, and perhaps necessary, technology to improve pandemic preparedness was described last year in an article that also mentioned the European HaDEA initiative. Then, too, the conclusion was that AI alone is not enough for that. This also requires better integration of data and analytics into the field of epidemiology.
