Optimizing Semantic Interoperability in Health Insurance: A Quantitative Approach

Abstract

Fragmented health insurance datasets remain a significant barrier to effective regulation, fraud detection, and national health analytics in low- and middle-income countries (LMICs). Variability in data structures, enrollee category labels, demographic fields, and provider identifiers across Health Maintenance Organisations (HMOs) undermines interoperability efforts and prevents the development of a unified national beneficiary database. This study presents a quantitative evaluation of an ontology-driven harmonisation pipeline designed to address these challenges by improving the consistency, semantic coherence, and structural alignment of administrative insurance records. A total of 216,080 enrollee records from multiple HMOs were analysed before and after harmonisation. Conflicts were categorised as semantic, syntactic, demographic, or schematic. The Enhanced Health Insurance Ontology (EHIO) guided standardisation, mapping, and semantic conflict resolution. Confusion matrices were constructed to evaluate correct mappings (true positives), incorrect mappings (false positives), missed harmonisable values (false negatives), and correctly rejected mismatches (true negatives). Performance metrics included precision, recall, F1-score, and accuracy. The harmonisation pipeline produced clear performance gains, with precision improving from 0.62 to 0.89, recall from 0.57 to 0.83, the F1-score from 0.59 to 0.85, and overall accuracy from 0.68 to 0.91. Paired t-test analysis confirmed that these improvements were statistically significant, indicating that the observed gains were meaningful rather than due to chance (p < .05). Additional benefits included a 34% reduction in duplicate enrolments, 29% correction of demographic inconsistencies, 92% resolution of semantic conflicts, and a 37% improvement in mapping uniformity. These results demonstrate that ontology-driven harmonisation provides a measurable, replicable, and scalable method for improving the quality of health insurance datasets in Low- and Middle-Income Countries (LMICs). The approach offers strong potential to strengthen national beneficiary registries and support digital health reforms aimed at achieving Universal Health Coverage (UHC).

DOI: https://doi.org/10.5281/zenodo.19097387