Healthcare information is by design spread across a variety of data sources. Medical documents generated for the same patients and the same diagnosis can be placed in different databases and are not necessarily associated with one another.

Semantic integration is the process of establishing links between documents sharing similar meaning. A myocardium and heart are examples of terms which have something in common that is not based on any syntax or spelling. Classical indexing methods using syntax only would therefore not consider any of those terms relevant to the other. Semantic-based methods on the other hand will infer that both terms are actually two facets of the same concept: heart disease.

We developed Nhumi Semantics to perform semantic integration over healthcare documents. Nhumi Semantics uses existing terminologies such as SNOMED CT, ICD-10, MeSH, etc. as knowledge bases modeling how medical terms and concepts relate to each other. Nhumi Semantics indexes any unstructured and (semi-)structured text according to its meaning and creates semantic links that can be used for search, patient’s overview creation, data aggregation, etc..