AI-enabled targeting of public health interventions through dynamic characterisation of the environment

The health benefits of diet and physical activity are well established. However, the environment influences these behaviours and hence health risk. Previous research has examined the role of the environment using people's addresses but this misses where people go, when they go, and what they do. This means that the interplay between environmental exposures and personal behaviours is missed.

We address this gap using existing timestamped measures of physical activity, location and diet from a large population-based cohort of over 10,000 adults in England. Using AI methods including autonomous agents acting on a knowledge graph, we will combine location and geospatial data to quantify personal dynamic environmental exposures to support epidemiological analysis of the cohort.

Results will be relevant to local authorities, including more detailed understanding of how people use space and the duration and timing of health-related exposures. Long term, this research could facilitate real-time nudges of health behaviours using smartphones.