Semantic Interoperability in AI-Supported MCI-to-AD Prediction: A Replicable FHIR-to-OMOP Framework
Semantic Interoperability in AI-Supported MCI-to-AD Prediction: A Replicable FHIR-to-OMOP Framework

Summary/Abstract: Semantic interoperability is a fundamental requirement for predictive models based on artificial intelligence in degenerative neurological diseases, particularly in the early stages of mild cognitive impairment (MCI) leading to Alzheimer‘s disease (AD). Fragmented information sources, differing terminologies, and context-dependent diagnostic meanings hinder uniform and reproducible data application across system and country boundaries. This scientific analysis presents a compact, AI-based framework for semantic interoperability based on HL7 Fast Healthcare Interoperability Resources and the Observational Medical Outcomes Partnership data model (OMOP-CDM). Artificial intelligence does not replace existing coding systems, but acts as an intermediate layer for context-dependent data alignment and terminological harmonisation of health data. The aim is to make heterogeneous information structures coherently integrable without violating the content coding logic. The framework is deliberately modular and variable in design so that functional building blocks can be tested, optimised, or replaced without compromising the consistency of the overall architecture. Using the prediction of the transition from mild cognitive impairment (MCI) to Alzheimer‘s dementia (AD) as an example, it is shown how AI contributes to content homogeneity, especially in early stages with unclear symptoms, where logic-based data harmonisation alone reaches its limits. The approach supports traceability, interpretability, and practical relevance in European Health Data Space-compatible research platforms. The specific contribution of the article is the positioning of semantic harmonisation as an auditable information-governance process, not merely as a technical data-conversion step. This clarification strengthens the distinction between AI-supported terminology alignment, predictive modelling, and clinical decision support.

• Details
• Contents