AI for public health disease surveillance

Artificial Intelligence (AI) promises to transform every industry, including public health. AI has the potential to play a crucial role in public health epidemiology and disease surveillance. Generative AI tools can help public health organisations by enhancing decision-making, data analysis, prediction, and public health communication processes, while improving operational efficiencies:

Decision support: AI could support disease surveillance by helping to triage cases and contacts based on symptoms, medical history, and risk factors to determine the urgency of outreach and recommending next best steps for the investigation. AI could improve surveillance data quality (e.g., suggesting record classification values based on disease-specific case definitions). And AI could optimise resource allocation and outbreak response planning, predicting the demand for healthcare services, hospital beds, vaccines and medical supplies, and channeling limited resources towards at-risk individuals and populations for greatest impact.

Analysis & Prediction: AI could enable better understanding of populations at risk and disease transmission dynamics by analysing and interpreting vast amounts of data from multiple sources to detect early signs of outbreaks, to monitor and predict pathogen spread, and to help identify sources of infection— all critical public health activities to mitigate transmission and protect populations from infectious diseases. This could include analysing unstructured clinical records, laboratory reports (e.g., monitoring a jurisdictional LIS for notifiable diseases to identify results that are reportable to public health authorities), environmental data (e.g., wastewater monitoring for pathogens of concern), and even news articles and social media for syndromic surveillance. By recognizing patterns and anomalies, AI systems could provide early warnings to public health officials and help them take proactive measures, targeting interventions and constrained resources where they are needed most. AI could be transformative in infectious disease epidemiologic exposure assessment, analysing free-text interview notes across multiple cases, contacts, and investigators and suggesting potential common source(s) of infection.

Communication: AI is the new UI! AI will enable citizen engagement in hyper-personalised ways. AI-enabled chatbots can generate responses that can deeply understand, anticipate, and respond to citizen issues. Dynamically-generated natural language disease/outbreak-specific questionnaires (e.g., food history, contact tracing) could be delivered in the client’s language of choice. AI could also support more effective public health messaging and communication by analysing social media trends, news articles, and online forums to gain insights into public sentiment and awareness regarding diseases, and tailoring public health messaging to combat misinformation while boosting dissemination of accurate information from reputable sources.

Operational efficiency: AI can help to address public health workforce challenges and address staff burnout by reducing administrative burden and improving the efficiency and effectiveness of overburdened personnel. AI can condense large volumes of structured and unstructured data into concise and informative synopses (e.g., generating case or outbreak summaries), saving investigator time and effort.

Sounds pretty magical, right? But we need some guardrails.

AI applications in public health must account for inherent concerns around privacy, security and bias. (a great example of AI bias— I generated the cyber-epi image at the top of this article using AI, and the first iterations it presented were all men; despite the fact that the field of epidemiology is dominated by women, the algorithms default to male scientists. And what the heck is the image at the top right?!)

Furthermore, the implications of AI hallucinations in health applications are far more grave than in other domains like sales or marketing. These tools must always be used in conjunction with human expertise and decision-making (“human-in-the-loop”), and remain firmly grounded in ethical frameworks that are still evolving.

The transformative potential of generative AI technology for public health cannot be ignored. AI capabilities are advancing quickly, and the field should?prepare to take advantage of these tools to bolster public heath practice.

I'd love to hear your thoughts in the comments below!