OBJECTIVES: The research agenda of algorithmic public health seeks to assess whether some typically human-driven efforts in health promotion can be replaced by AI-assisted processes. In order to assess the possibility of algorithmic approaches to public health, an assessment of the current quality of public health policy-making is needed. As a specific example, this research will focus on campaings to promote healthy lifestyles in the Czech Republic, proposing an alternative algorithmic model for personalized prevention. The guiding research question is "Do AI-assisted health promotion efforts lead to superior results as compared to human-driven strategies?"

METHODS: The research draws on the conceptual framing of Brownson's "evidence-based public health" (EBPH), scrutinizing the various principles it lays out. The findings will be contrasted with the potential that AI LLMs can produce when prompted to devise a public health messaging strategy based on a set of inputs. A group of independent evaluators will be asked to grade to what extent the two strategies meet the EBPH, being blinded to which alternative they are grading.

RESULTS: The preliminary findings suggest that human-driven (standard) approaches to health promotion efforts fail to live up to the criteria set by Brownson. Conversely, there is great potential for AI-driven approaches to achieve superior results. This will need to be tested using the experimental setup mentioned above, which is planned for September 2023. Partial results will be presented during the conference.

CONCLUSIONS: This conceptual paper seeks to bring about a new concept of algorithmic public health and test out its practical potential through the lens of health promotion strategies in the Czech Republic. The implications can extend to other countries and public health domains, depending on the results of further research efforts.