Identification of Ecological and Human Health Indicators Based on Geographic and Spatial Mapping

Addressing public health and environmental challenges in rapidly urbanizing areas requires identifying and analyzing relevant indicators. This study introduces an AI-powered framework for extracting and analyzing health and ecological indicators from biomedical literature. Using PubMed as a data source and Large Language Models, we processed 16,474 publications from 5 European regions related to the EU-funded OneAquaHealth project. We automated the identification of 41,377 unique georeferenced relationships between health and environmental indicators. Through dimensionality reduction and clustering, we elicited eight healthcare-focused clusters (e.g., chronic disease and mental health) and eight environmental-focused clusters (e.g., exposure to contaminants). Spatial patterns highlighted disparities in data density, with greater coverage in Western Europe and less in Africa and the Middle East. Our methodology demonstrates how extracting valuable, geographically based insights from unstructured biomedical text supports interdisciplinary research and helps decision-makers address health and environmental challenges.