INTRODUCTION:
public health institutions routinely publish large volumes of monitoring data in narrative pdf reports. Although these reports contain valuable information, their unstructured format limits large scale analysis and reduces their usefulness for evidence based policy decision making. This challenge is particularly evident in environmental health reporting, where recurring temporal patterns may remain obscured within descriptive documents.

METHODS:
this study presents an ai assisted workflow for transforming unstructured public health reports into structured and policy relevant datasets. publicly available monthly air quality monitoring reports covering a three year period were used as a case study. pdf text was extracted using python based tools, followed by controlled ai assisted parsing to identify explicitly stated indicators such as the number of days exceeding regulatory thresholds for pm10. ai was used solely for text extraction and data structuring, while interpretation and analysis were conducted by the author. extracted data were aggregated temporally to identify patterns relevant to public health policy.

RESULTS:
the structured dataset revealed a strong and consistent seasonal pattern, with the majority of exceedance days occurring during winter months and minimal exceedances during the remainder of the year. this temporal concentration was not readily apparent when reviewing individual reports but became clear after aggregation across time. the findings demonstrate how ai assisted structuring substantially increases the analytical value of routinely published health information without altering underlying data.

CONCLUSIONS:
ai assisted analysis of unstructured public health reports can strengthen health information systems by improving data usability and policy relevance. this approach supports more timely and targeted public health interventions and is transferable to other domains of public health reporting.