UNMIRI opens precision oncology APIs to design partners

UNMIRI said its precision oncology API platform is now available to design partners and developers, giving hospitals and software teams a cross-vendor way to normalize next-generation sequencing reports and ground clinical decisions in public sources. The launch targets a common oncology workflow problem: turning vendor-specific genomic reports into traceable data that can support therapy selection, trial matching, and prior authorization.

Why it matters: - Oncology data often arrives in different formats from different sequencing vendors, which makes interpretation uneven across clinics and software systems. - UNMIRI is positioning its platform as a vendor-neutral layer for genomics-aware decision support, trial matching, and prior authorization. - The company is targeting verifiable outputs in a field where a wrong variant interpretation can change treatment choices.

What happened: - UNMIRI LLC said its precision oncology software platform is available now to design partners and developers. - The platform is a set of APIs that converts next-generation sequencing reports from major labs into normalized clinical data with citations. - Access is open through synthetic-data sandboxes. - UNMIRI said the company remains pre-revenue and pre-pilot at this stage.

The details: - The platform parses reports from nine or more vendors into a single FHIR R4 Genomics representation. - The output is mCODE-compatible for registry and EHR pipelines. - The system normalizes variants with HGVS standards. - It also captures biomarkers including TMB, MSI, HRD and PD-L1, plus companion-diagnostic flags. - A genomics-aware decision support API returns drug-variant pharmacogenomics, variant-specific therapy options graded by the public AMP/ASCO/CAP 2017 tiers, hereditary-cancer triggers, and trial matches. - A variant-grounded trial-matching API is built for organizations that match patients to studies. - A prior-authorization decision engine grounds coverage decisions in FDA labels and the Medicare local coverage determinations it indexes. - The architecture uses a typed Neo4j knowledge graph to resolve variant identity against standardized identifiers before retrieval or decision steps. - Final clinical output is rendered by deterministic templates, not language models. - UNMIRI said language models are limited to extraction edge cases and never see patient data in prompts. - Every recommendation traces to public sources including CIViC, ClinVar, ClinicalTrials.gov, openFDA, and CPIC. - The platform runs on HIPAA-ready AWS infrastructure under an active business associate agreement. - Patient data is processed in memory and not retained. - UNMIRI said the decision support API is designed to meet FDA Non-Device Clinical Decision Support criteria under the 21st Century Cures Act. - The design includes structured inputs, options rather than directives, and a cited basis a clinician can review independently.

Between the lines: - The launch is aimed at trust and auditability as much as workflow speed. - UNMIRI is leaning on exact variant identity and deterministic output to avoid errors that can come from approximate retrieval or generic AI responses. - The company is framing its product as infrastructure for other healthcare software companies, biotech medical affairs teams, and clinicians rather than as a consumer-facing tool.

What’s next: - UNMIRI is onboarding design partners now. - The company said the platform is built and deployed, with broader use expected to start from synthetic-data testing and partner feedback. - UNMIRI was founded in 2023 and is based in Langhorne, Pennsylvania. - The two-person team says the business started after a family caregiving experience exposed how fragmented oncology reporting had become. - More information is available at UNMIRI.