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

Area of Science:

• Public Health
• Environmental Science
• Computational Biology

Background:

• Urbanization presents complex public health and environmental challenges.
• Identifying and analyzing relevant health and ecological indicators is crucial for addressing these challenges.
• Biomedical literature contains valuable, yet unstructured, data relevant to these indicators.

Purpose of the Study:

• To develop and apply an AI-powered framework for extracting and analyzing health and ecological indicators from biomedical literature.
• To identify georeferenced relationships between health and environmental indicators within European regions.
• To support interdisciplinary research and decision-making for urban public health and environmental management.

Main Methods:

• Utilized PubMed as a data source, processing 16,474 publications.
• Employed Large Language Models (LLMs) for automated information extraction.
• Applied dimensionality reduction and clustering techniques to analyze extracted data.

Main Results:

• Identified 41,377 unique georeferenced relationships between health and environmental indicators.
• Elicited eight healthcare-focused clusters (e.g., chronic disease, mental health) and eight environmental-focused clusters (e.g., contaminant exposure).
• Revealed spatial disparities in data density across European regions, with less coverage in Africa and the Middle East.

Conclusions:

• The AI framework effectively extracts geographically based insights from unstructured biomedical text.
• This methodology supports interdisciplinary research by linking health and environmental factors.
• The findings provide valuable data for decision-makers tackling urban public health and environmental issues.