How It Works

How evidence-gating works.

Neurona 28 does not treat your input as a prompt for generative expansion. It treats it as a bounded evidence set.

Four steps. Evidence at every one.

01 — Review

The professional reviews the patient's medical record in their EHR — progress notes, lab results, imaging reports, encounter documentation. This is where human expertise begins.

02 — Input

The professional types clinical findings directly into Neurona 28 in the language they naturally use — including the Spanish-English code-switching characteristic of Puerto Rico's clinical workforce. No EHR connection. Deliberate, not a limitation.

03 — Validate

The platform parses input through its Golden Record-governed clinical logic, resolves mixed-language descriptions, maps findings to V28-compliant ICD-10 codes, and applies evidence-gating to confirm every suggested HCC is explicitly supported by documented input. Codes without sufficient evidence are not suggested — they are flagged.

04 — Output

Validated V28 coding suggestions with full evidence traceability. Each code linked to the specific clinical finding that supports it. Structured with MEAT documentation for RADV defensibility from the moment it is generated. Gaps flagged with non-leading physician queries.

Core Concept

What is evidence-gating?

Evidence-gating means the system will not suggest a code without documented clinical support. If the clinical input does not contain sufficient evidence for a diagnosis, the system stops. It does not infer. It does not assume. It does not fill the gap. It flags the gap, returns to the user, and — when appropriate — generates a non-leading physician query to request documentation clarification.

This is the opposite of how generic large language models work. A general-purpose AI will generate a plausible answer. Neurona 28 generates a defensible answer — or it generates nothing.

Synthetic example — not real patient data

See it in action

CKD Staging Gap

Synthetic example — not real patient data

Input

Paciente 67 años, masculino. Assessment: DM2, HTN, CKD. Lab: eGFR 38 mL/min.

What N28 does

Maps documented findings. Identifies CKD is present but staging is unspecified. eGFR of 38 suggests Stage 3b but the physician has not documented the stage in the A/P.

What N28 refuses to do

Does not assume CKD is diabetic without explicit linkage documentation. Does not assign a stage without physician documentation.

Output

“Flags the staging gap. Generates a non-leading physician query: 'The lab data suggests the patient's kidney function may correspond to a specific CKD stage. Would you be able to document the CKD stage in the assessment?'”

Why it matters

Coding unspecified CKD (N18.9) vs. Stage 3b (N18.32) has significant risk adjustment and care management implications. The evidence gate prevents the coder from assuming a stage that the physician has not documented.

Join the first cohort.

We are onboarding a select group of risk adjustment professionals in Puerto Rico. Request early access to be considered.

Evidence-gated — no guessed codes
V28-native from day one
Built for Puerto Rico's clinical workflow