Agentic AI for public health contact tracing & beyond

Public health practice and field epidemiology, long hampered by antiquated technology solutions, are undergoing a major transformation in many health departments around the world, in part because of digital advancements helping to drive efficiency and accuracy in areas like population disease modeling and surveillance. The application of artificial intelligence (AI) emerged as a powerful public health tool during the COVID-19 pandemic due to the ability of AI models to analyze and interpret large public health datasets at scale, enabling evidence-informed policy development and targeted program implementation in real time.?

In the technology sector, sometimes it feels like we’re already living in the future– advancing quickly from predictive to generative, the world of AI has been evolving at an unprecedented pace. Just when we were getting comfortable wrapping our heads around the transformative potential that AI holds for public health, there’s a new breakthrough on the AI scene: agentic AI.?

AI agents are not mere chatbots; they are fundamentally shifting the paradigm of AI capabilities. We find ourselves thinking of the omnipresent Computer on Star Trek (TNG, of course, because that’s still the best Star Trek): a helpful, always available technology assistant with extensive depths of knowledge that is capable of reasoning.?

Typical AI chatbots use predefined rules to answer simple natural language questions; AI agents can process vast amounts of data, make decisions, and perform complex, multi-step tasks. Where earlier forms of AI operate on simple tasks with predefined actions and summarization capabilities, agentic AI autonomously initiates a sequence of actions, can anticipate next steps, and come up with creative solutions in response to a request. AI agents can understand the nuances of different business processes, data formats, and industry-specific requirements, and they have sophisticated tools like reasoning engines and safety guardrails to ensure controlled and predictable behaviour. And AI agents learn continuously from interactions, refining responses over time while adhering to organizational permissions and policies.?

Some degree of autonomy is key to agentic AI. AI agents operate autonomously, or semi-autonomously, depending on their purpose or complexity; AI agents for many use cases work best when paired with humans, supporting decision-making while operating within predefined guardrails. This enhanced autonomy will transform the application of AI in public health and promises to revolutionize the way that public health practitioners solve complex population health challenges.?

Take for example contact tracing- an essential but resource-intensive aspect of public health response that informs population health practitioners about the epidemiology of infectious and communicable diseases in their community, allowing them to take targeted action to prevent and interrupt spread. During many outbreaks, the potential scale of pathogen transmission and sheer volume of investigative data may make it difficult to comprehensively assess acquisition and transmission exposures and determine the source(s) of infection. This is necessary to interrupt the chain of transmission to prevent ongoing spread and mitigate future outbreaks, and means that far too many outbreaks remain unsolved. Novel potential sources of infection often emerge during case investigations, and need to be rapidly vetted for further investigation. Agentic AI could be transformative in infectious disease epidemiologic exposure assessment, analyzing survey responses and free-text interview notes across numerous cases, contacts, and investigators to determine potential exposure sources that warrant further investigation. AI agents could further streamline case follow-up processes by automating routine activities like administering contact tracing questionnaires or quarantine monitoring, freeing up human resources to dive deeper into the data to identify common sources of infection.?

Other opportunities for agentic AI to support contact tracing and public health case investigation include automated compliance/quality assurance checks on case files, including dynamic reminders to investigators to document missing information that’s required to inform outbreak investigation and response. Agentic AI could also automate public health communication by synthesizing and generating personalized disease or outbreak-specific fact sheets and health guidance for investigators to share with cases/contacts, right in the flow of their work. AI agents could handle inquiries from the public 24/7 with personalized outreach for health promotion, recommending appropriate health programs and services tailored to local language and context, lifestyle, and literacy levels.

Of course, there are privacy, security, transparency, and equity implications with the application of any new technology, particularly in health. The aim should be to stay aligned with human values, with AI agents augmenting human capabilities and capacity, not replacing them.?

Public health is sorely lacking the resilience it will need to respond to future pandemics or other health emergencies. Agentic AI can help to address public health resource and workforce challenges and staff burnout by reducing administrative burdens and improving the efficiency and effectiveness of overworked personnel. It can also enable delivery of services in a digitally native manner, which can increase engagement with the community.?

AI agents are not going to replace most human jobs any time soon, particularly not in public health. The public health workforce has not recovered from the impact of the COVID-19 pandemic, leaving a cohort of exhausted staff who are struggling to do more with less. With AI agents on board, labour can be reallocated, shifting the public health workforce to other high-value tasks and programs.?

High tech digital interventions to enable more high quality analog public health interactions? Make it so.

~ Karen Hay is a Digital Health Transformation Executive/Health Industry Advisor and Roksolana C. is a Principal Solution Engineer at Salesforce . Both have backgrounds in epidemiology and get far too excited thinking about how technology can support public health practice.