The Geo Health space has strong coverage of drugs, vaccines, conditions, devices, and clinical pathways — but several structurally important entity types are absent. The 10 gaps below were identified by auditing the space's type ontology and running searches across key health domains. Each missing entity would anchor a meaningful cluster of related knowledge that currently has no typed home.
The space models biological and clinical factors richly but has no typed entity for upstream social drivers — income, education, housing, food access — that the WHO estimates account for 30–55% of health outcomes, making it impossible to link conditions or treatments back to the root causes that determine who gets sick in the first place.
While the space has a "Study category" type and scattered claims referencing individual trials, there is no first-class Clinical Trial entity to structured-link a drug or treatment to its phase, sponsor, NCT identifier, primary endpoint, and outcome — the core provenance chain that distinguishes evidence-based medicine from anecdote.
The space has Lab test and Health metric types but no Biomarker entity — a measurable biological indicator (e.g., HbA1c, CRP, BRCA1 expression) that bridges conditions, diagnostics, and drug targets; without it, the graph cannot represent the molecular logic that connects a disease's mechanism to how it is detected and monitored.
With sequencing now embedded in news stories and claims across the space, a first-class Genetic Variant entity (carrying gene, chromosome position, allele, and clinical significance) is essential to link disease risk, pharmacogenomics, and hereditary conditions in a queryable, structured way rather than burying them in free-text claims.
The space references trachoma, Ebola, and tropical diseases in claims but lacks an NTD entity type — a structured record linking the pathogen, endemic geography, affected population, and elimination status — leaving a blind spot for the 20+ WHO-classified diseases that disproportionately burden low-income and African populations, including Nigeria.
The space has a type stub labelled "Antimicrobial resistance mechanism" but it is typed only as "Type" with no instances, properties, or relations — it needs to become a populated entity class linking specific resistance genes (e.g., MCR-1, NDM-1), the bacteria carrying them, and the antibiotics rendered ineffective, which is the foundational data structure for any AMR surveillance graph.
End-of-life care appears only in news stories and ethical questions, but the space has no structured Palliative Care Protocol entity — leaving no machine-readable way to link symptom management goals, medication plans, setting (home vs hospital), and cultural or religious considerations that are especially critical in under-resourced healthcare contexts across sub-Saharan Africa.
The space can model individual drugs in detail but has no Drug–Drug Interaction entity to capture the pair, the mechanism (e.g., CYP450 inhibition), the severity, and the clinical consequence — a critical gap given that the space already tracks antimalarials (which have notorious interactions with common co-medications like artemether-lumefantrine and rifampicin).
Herbal appears only as a Supplement category, but there is no entity type for Traditional Medicine Practices — the structured therapies of Ayurveda, TCM, Yoruba ethnobotany, and other systems — despite the WHO estimating 80% of the global population relies on them as primary care, and despite their direct relevance to drug interaction risks and public health communication in Africa.
Disease burden and chronic disease management exist as topic tags, but there is no Comorbidity Pair entity — a typed, directed relation between two co-occurring conditions (e.g., HIV + TB, diabetes + hypertension, sickle cell + malaria) with prevalence data and shared mechanism — which is precisely the structure needed to model the compound disease burden driving the majority of mortality in Nigeria and Africa.