Public development demo Founder-built project · development and evaluation stage

Reviewable clinical-coding suggestions for human coders.

EnayaCode explores how AI can extract candidate diagnoses and procedures and present ICD-10-AM / ACHI suggestions with supporting text for coder review. It does not make autonomous coding, billing or clinical decisions.

Illustrative synthetic example · not a performance result
Source clinical text
Male, 62. Presented with chest pain. Diagnosed with acute myocardial infarction. Underwent coronary artery stent insertion via PCI on day 1. Discharged on day 3, stable.
ICD-10-AM I21.4 Acute subendocardial myocardial infarction example
ACHI 38306-00 Percutaneous insertion of coronary artery stent example
ACS 0940 Acute coronary syndrome · ACS 0940 sequencing rule example
Designed for review logging Synthetic demonstration
This site is a public development demo of EnayaCode. Demonstration cases, records, people, organisations and any figures shown are synthetic or illustrative. Nothing here implies measured accuracy, regulatory approval, commercial availability or a support commitment.
Human approval required Arabic and English interface Tenant-scoped data model Development-stage software
Capabilities under evaluation

A structured workflow from source text to coder review.

These are software capabilities, not claims of measured accuracy, fewer rejections or regulatory approval.

Candidate-code generation

The workflow extracts clinical concepts, proposes candidate codes and can restrict suggestions to entries found in configured code tables. A coder must check every result.

Review evidence

Suggestions can include a rationale, confidence breakdown and source-text anchor so a reviewer can inspect why a code was proposed.

Bilingual end-to-end

The interface supports Arabic (RTL) and English (LTR). Language support does not imply clinical validation in a particular country or health system.

Tenant-scoped architecture

The application includes organisation and role scoping. This is an architectural feature, not a security certification or assurance for production health data.

Coding tiers

Planned coding tiers

Tier concepts under development, not commercial plans.

Tier 1

Standard

Two-stage pipeline: AI extraction, then selection restricted to codes checked against reference code tables.

  • Deterministic code-table checks on suggestions
  • Confidence and rationale shown with suggestions
  • Source-quote highlighting
  • Excel + JSON export
Tier 2 · evaluation

Advanced

On-demand second opinion: an independent blind re-code compared and adjudicated against the original.

  • Independent blind re-code + adjudication
  • Confidence calibration from reviewer outcomes is under evaluation
  • Designed to fail closed on low confidence
  • Detailed audit dashboard
Tier 3 · concept

Premium

Live reference retrieval woven into coding for the highest-stakes claims.

  • Designed to ground suggestions in clinical reference data
  • ACS sequencing rules support is planned
  • Aimed at demanding audit scenarios
  • Awaiting licensed reference data partnership
Workflow

From source file to a coder-reviewed export, in four steps

1
Upload the document

Drop in the discharge summary or operative report as PDF or DOCX, or paste the text.

2
AI analysis

The engine extracts diagnoses and procedures and proposes candidate codes with rationales.

3
Coder review

Accept, edit, or reject each code in a two-pane workbench. Actions are captured in an audit trail.

4
Export the claim

Export reviewed information as Excel or JSON for further checking and integration. The export is not automatically a payer-ready claim.

Intended users

Who the workflow is being designed for

Hospitals and Clinics

Teams evaluating structured, human-reviewed coding workflows.

Clinical Coders

An AI assistant proposes. You approve, edit, or reject.

Insurance Payers

Reviewers evaluating the provenance and disposition of proposed codes.

Access is restricted to authorised evaluation users.

Do not upload real patient or confidential data without an approved evaluation agreement and data-handling controls.

Authorised sign in