Six modules for value-based primary care — including our Referral Review Engine, which reduces total referrals 40%+, appropriately.
Each module adds value on its own. Adopt one for immediate impact or all six for closed-loop accountability — across your tech stack or Alma's full platform.
Memory-based specialist routing eats coordinator and clinician time. The work scales linearly with volume — there is no leverage. Conditions that should be PCP-managed are routinely sent out. Defensibility is anecdotal. Audit trails are thin.
Time spent looking up specialist preferences, deciding pathways, and writing justifications. Multiplied by every referral, every PCP, every day.
Industry-reported share of specialty referrals that are PCP-manageable. Uncomplicated HTN, well-controlled DM, benign skin lesions sent out unnecessarily.
Direct: specialist visit, downstream imaging, repeated workup. Indirect: patient delay, narrative friction, network leakage.
Validated production data shows a sustained reduction in total specialist referrals, appropriately, across multiple production deployments.
See the methodology →Optional preferred-network integration. When configured, the engine reports appropriate-referral rate alongside out-of-network leakage rate — both as bundled metrics for plan dashboards and ACO reporting.
Share of referrals routed in concordance with cited clinical guidelines. The directional operating signal that the engine is doing the work it was designed to do.
Share of referrals routed outside your preferred-network configuration. The engine respects in-network preferences while preserving guideline concordance — and reports the gap when one occurs.
Cost avoidance through inappropriate-referral prevention, with an immutable record of guideline source, engine version, and timestamp on every recommendation.
For MA plansAligns with CMS innovation models that reward primary-care-managed pathways, defensible referral patterns, and equity-aware operations.
For ACOsCompresses referral routing to under two minutes. Two-input simplicity. Drop-in for eClinicalWorks, Athena, NextGen workflows.
For providersA complete platform for healthcare plans and risk-bearing provider groups under value-based contracts: front-door triage through downstream analytics. Each module addresses a distinct operational layer — deployable independently or as an integrated stack.
No. The Referral Review engine is decision support, not autonomous decisioning. Clinician override authority is preserved on every non-safety decision. The 39 hardcoded patient-safety escalations cannot be bypassed by any user — that boundary protects the patient, not Alma.
Validated production measurement across multiple production deployments. The reduction is observed against pre-deployment routing patterns and active operations. Methodology — referrals per thousand measured before and after deployment, with PCP-LED routing rate tracked in production — is documented. Cohort definitions, comparator structure, and validator credentials are available on request under NDA.
Drop-in for eClinicalWorks, Athena, and NextGen referral workflows. Output maps to standard referral status fields (Pending / Approved / Routed-to-PCP / Hold). FHIR R4 interoperability is a design principle. Epic and Cerner via partner connectors where in place.
Thirty days. Joint scoping of success criteria up front — typically: ≥30% reduction in total referrals, appropriately, override rate < 15%, zero missed safety escalations, ≥99% engine uptime. We bring clinical configuration support and full EHR integration — our team handles the technical lift end-to-end. The platform can also operate as a standalone service: ingest referral indication and specialty data in bulk, produce reports and routing recommendations, and deliver them to any person or department the client designates.
HIPAA-aligned design. SOC 2 Type II–attested infrastructure (Supabase, Vercel). Append-only audit logs. Field-level diff tracking via Postgres triggers. Role-based access. No SSN capture. The engine generates an immutable recommendation record with engine version, guideline version, and timestamp on every call.
A 30-minute working session on a sample of your referrals. We'll model your specific time and dollar savings — for context, a 10,000-patient panel sees modeled annual cost avoidance of $5.76–7.14M and recovers approximately 4,500 coordinator-hours.
Request a demo →