Search for a command to run...
In the European Union, all medications must include a paper-based medication leaflet providing essential information. However, challenges with paper leaflets have driven efforts toward digitization and standardization. This paper presents an automated extraction, digitization, and standardization pipeline for medication leaflet information, integrated into a mobile medication management application. The automated extraction is using an LLM-based question-answer pipeline. Extracted information was standardized through template-based mapping to FHIR resources, where LLM-structured content was integrated into profile-aligned templates and validated using the $validate operation. To assess pipeline performance, ten randomly selected medication leaflets were evaluated, with 36 extracted fields per leaflet scored binary for correctness, completeness, and format adherence. Errors were categorized into four types: assignment errors (separating coherent information), missing information, interpretation errors (handling ambiguous content), and false information (hallucination). Mean scores were calculated per question set to identify extraction patterns. Fields with clearly delimited information—such as indications, excipient ingredients, and dose form—achieved mean scores of 0.9 or higher, with 20 of 36 fields demonstrating perfect accuracy. However, complex nested sections showed substantially lower performance: precautions scored 0.20 (predominantly assignment errors), interactions 0.30 (assignment errors), and undesirable effects 0.40 (primarily missing information). These recurring errors indicate the pipeline requires refinement for heterogeneous structural patterns in nested information. Structural inconsistencies in information hierarchies most significantly impacted extraction accuracy for these clinical use cases. By addressing these challenges, this approach opens new pathways in healthcare information management for mobile health developers and healthcare IT professionals. The mobile application could potentially be used by patients taking different medications and aims to improve medication literacy and accessibility. This work combines information extraction and standardization processes, laying potential groundwork for extending the approach to other healthcare areas. Practical benefits include enhanced searchability, multi-system interoperability, and improved patient access to structured medication information through mobile applications.