Public Health Data Harmonization Frameworks

Abstract

Public health data harmonisation -- the systematic alignment of heterogeneous data sources from multiple jurisdictions into shared analytical representations -- is a precondition for coordinated cross-border public health surveillance, comparative effectiveness research, and evidence-based international health policy. The challenge is substantial: national surveillance systems use different case definitions, reporting cadences, demographic stratifications, and data quality standards that reflect divergent historical development rather than deliberate variation. Cross-border public health events -- pandemic response, antimicrobial resistance monitoring, cross-border disease outbreaks, climate-driven health impact assessment -- require data harmonisation infrastructure that can operate in near-real-time rather than through months-long retrospective harmonisation exercises. We evaluated five public health data harmonisation frameworks applied to 32 European countries' surveillance systems covering 18 major disease categories and public health indicators between 2018 and 2024. Frameworks ranged from manual retrospective harmonisation through ML-based automated semantic harmonisation with federated analytics infrastructure. Performance was benchmarked using our Public Health Data Harmonisation Effectiveness Score (PHDHES), integrating harmonisation completeness, temporal responsiveness, cross-border analytical capability, privacy compliance, and implementation sustainability. The ML-based automated harmonisation framework achieved the highest PHDHES (0.914), reducing time from data receipt to harmonised analytical dataset from 42 days to 2 days and enabling cross-border outbreak analyses within 72 hours of signal detection versus 3-4 weeks under conventional manual harmonisation.