Presentation (CTI)

OBJECTIVES: Health economic models and reports often need to be repeatedly adapted as per the requirements of different regions and countries. Model and report adaptation has traditionally required considerable manual effort or rule-based automation through tools like VBA. These conventional approaches are deterministic, lack adaptability, and cannot be classified as true artificial intelligence. This study aims to demonstrate an agentic AI-based implementation that autonomously adapts Excel-based health economic models and corresponding Word-based reports for country-specific use without requiring any manual or rule-based pre-configuration.
METHODS: The implementation operates in two modules. In the first module, the user uploads a global Excel model and a list of parameters tailored to the target country. The AI system scans all sheets, identifies relevant tables and cells based on structural cues (titles, labels, table boundaries), and replaces default values with country-specific ones. It then generates a fully adapted Excel model. In the second module, the user provides a global Word report and the adapted Excel model. The AI reviews each section of the global report, automatically compares the data with Excel model, reasons over the changes, interprets them and writes them in the report section by section. The rewritten text does not merely replace data but also updates interpretation without any placeholder abstracted text. Notably, no templating, pre-defined rules, or VBA scripting is used, demonstrating true generative and autonomous functionality.
RESULTS: The tool successfully adapted models and reports across multiple country settings without any time and labour-intensive manual tagging or rule-based configuration. It consistently identified and replaced model parameters and report content with contextual accuracy.
CONCLUSIONS: This tool exemplifies the distinction between traditional automation and AI in HEOR workflows. As AI adoption grows, clarity in distinguishing true AI capabilities from automation is vital for credibility and innovation in health technology assessment.