Prior Authorization Workflow
This document describes the end-to-end workflow for processing prior authorization requests through the Najeeb AI Engine. It covers the complete lifecycle from data synchronization through request submission, AI-powered adjudication, and final decision handling.
Workflow Overview
The Prior Authorization workflow follows an asynchronous processing model. Your core system submits a request via the API, the Najeeb AI Engine processes it, and the final decision is delivered via a webhook. A reviewer then updates the authorization status, which records it in the transaction history.
Prior Authorization / Standard Processes preauth requests and validates data against predefined entities for policy holders, policies, members, and medical providers. Failure to sync any of these entities results in errors and prevents creating the request.
Prior Authorization / Lightweight Processes preauth requests without validating predefined entities. Najeeb does not check for the existence of policy holders, policies, members, or medical providers, and automatically creates them while validating the request data.
Phase 1: Data Synchronization - Required only for Standard Preauth
Before submitting any prior authorization request, all reference data must be synchronized with your Najeeb instance. Missing data will result in 404 Not Found errors at submission time.
Required Entities
| Order | Entity | API Reference | Why It's Needed |
|---|---|---|---|
| 1 | Policy Holder | POST /policy-holder/create | Parent entity for all policies |
| 2 | Policy | POST /policy/create | Defines benefits, sub-benefits, and coverage rules |
| 3 | Member | POST /policy/patients/add | The patient must be registered and assigned to a policy |
| 4 | Medical Provider | POST /medical-provider/create | The provider's CCHI number must exist in the system |
| 5 | Webhook | POST /webhook/create | Required to receive asynchronous AI decisions |
Failure to sync any of these entities will result in errors such as E_EPH_01 (Policy holder not found), E_EPA_014 (Policy not found), E_EPA_01 (Patient not found), or E_EMP_01 (Medical provider not found).
Phase 2: Request Submission & AI Processing
Once reference data is synchronized, your core system can submit prior authorization requests. The processing is asynchronous: the API returns an acknowledgment immediately, and the AI Engine delivers the final decision via webhook.
What the AI Engine Evaluates
During processing, the Najeeb AI Engine performs the following checks:
- Diagnosis validation -- verifies ICD-10 codes against the Electronic Code List
- Policy check -- validates member enrollment, and coverage rules
- Medical necessity -- evaluates service-diagnosis consistency
- Price list validation -- compares claimed amounts against the configured price list (only if the Price List Module has been activated and synced beforehand)
- Patient history -- reviews the preceding year of approved authorizations (when
with_patient_history: true) - Fraud detection -- identifies duplicate requests, suspicious patterns, and inconsistencies
- Pharmacy (PBM) -- for pharmaceutical items, validates drug codes against the SFDA registry
Webhook Decision Payload
The AI Engine delivers results with item-level decisions, each containing:
decision-- the status for each item (ACCEPTED,REJECTED,PARTIALLY_ACCEPTED)reason_code-- array of detailed reasons with Najeeb AI codes and NPHIES-mapped codesnajeeb_ai_alert-- system-generated flags or observationsnajeeb_ai_score-- AI confidence scoreapproved_quantity/approved_amount-- adjusted valuespbm_drug_info-- pharmaceutical details (for drug items)
Phase 3: Review & Update
After receiving the AI decision via webhook, a reviewer in your system must call the Update Prior Authorization API to finalize the status.
You must call the Update Prior Authorization endpoint to set the final status. This is what records the authorization in the transaction history. Without this call, the prior authorization remains in a pending state and is not reflected in historical records.