AI Medical Evaluator sits inside the EHR and reviews every encounter through the payer's lens before the clinician saves. Documentation gaps, coding errors, medical-necessity failures, and medication-safety issues — caught in real time, while the patient is still in the encounter.
Claims are not denied because care was wrong. They are denied because the documentation did not establish medical necessity, coding specificity, or NPHIES compliance in the language payers require. These failures originate at the point of documentation — and that is where they must be caught.
First-pass rejections driven by symptom codes as principal diagnosis and missing secondary diagnoses.
Insufficient documentation of medical necessity creates downstream denial exposure that compounds across volume.
Retrospective CDI queries, resubmissions, and manual coding corrections consume hours per denied claim.
Nearly a third of submitted claims require correction before a payer will accept them.
Three steps that fit into the existing clinical workflow. No EHR-side installation. No workflow disruption. The clinician documents as usual — AI Medical Evaluator adds a second set of eyes, trained on the payer's rulebook.
The extension reads every field in the encounter form — diagnoses, medications, vitals, orders, procedures — using field-type classification that works across any EHR layout.
Deterministic rules and AI models run simultaneously: drug interactions, renal dosing, vital-range and computed indicators, ICD-10-AM specificity, NPHIES validation, medical necessity, and a 0–100 denial-risk score — with AI-drafted content clearly tagged for verification.
Findings surface in a side panel with severity levels, evidence citations, and one-click corrections. The clinician reviews each flag and saves documentation that withstands payer scrutiny.
Beyond flagging documentation gaps, the platform carries a working clinical toolset — calculators, code lookup, denial prediction, and bilingual review — so the clinician never leaves the encounter to answer a question.
CrCl, CKD-EPI eGFR, MELD-Na, CURB-65, CHA₂DS₂-VASc, Wells, GCS, Child-Pugh, opioid MME, pediatric dosing, infusion rates — every result shows the interpretation band and its source.
A 0–100 claim-denial score from eight documentation factors — symptom principal codes, missing necessity links, unlinked medications — with a factor-by-factor breakdown and fix-first recommendation.
Renal dosing by eGFR band, allergy cross-reactivity (penicillin → cephalosporin), pregnancy and pediatric checks, and a 300+ pair severity-graded interaction checker — contraindicated to mild.
Search ICD-10-AM, SBS, DRG, and the SFDA drug registry (8,756 drugs) directly in the panel — no network round-trip, click-to-copy codes while documenting.
Paste or type the visit narrative and review structured S/O/A/P sections ready to inject into the form — with voice commands ("accept all", "set temp 38.5") and Arabic numeral support. Live dictation capture is in rework.
The review interface switches to Arabic with RTL mirroring across panels, flags, and actions. Arabic clinical terminology is recognized in the form fields and interpreted in the review prompts themselves.
When the AI synthesizes exam narrative from vitals and findings, every value is tagged "AI-drafted — verify on examination" on the card and on the field. Advisory text is structurally blocked from injection.
Command palette (Ctrl+K), severity-colored field overlays with apply/re-check tooltips, A/R/N field shortcuts, critical-first sorting, and JSON / CSV / print exports of any review.
Validation history compares any two reviews — findings, criticals, warnings — so documentation improvement is visible as teams adapt. PHI-safe audit trail for compliance.
AI Medical Evaluator adapts its findings to the reviewer's role and workflow stage. Each stakeholder sees the issues that matter to their function — with the right level of detail and actionability.
Real-time flags while documenting: unspecified codes, missing diagnoses, medication-safety issues, and documentation gaps.
Verify that the documentation establishes medical necessity for every service rendered, per NPHIES requirements.
Catch NPHIES rejection risks before submission: missing fields, format errors, code mismatches, and claim-level validation.
Ensure the most specific ICD-10-AM code supported by documentation, correct DRG assignment, and complete capture of comorbidities.
Identify the root cause of rejections and denials, generate corrective documentation, and support appeal narratives with evidence.
Run audit simulations across encounter batches, identify compliance gaps, and generate defensibility reports with source citations.
AI processing runs on a governed server over an authenticated, encrypted channel. The browser extension handles DOM scanning and the review UI — provider API keys and prompt routing never leave the server. Exports and analytics are PHI-scrubbed before leaving the review session.
All extension–server traffic is JWT-authenticated and encrypted. No API keys are stored in the browser.
Saudi IDs, Iqama numbers, MRNs, phone numbers, and patient names are detected by the server-side PHI guard on every inbound request, and scrubbed from exports and analytics before anything leaves the review session.
Analytics store only metadata, hashes, and counts — no patient-identifiable information ever enters the analytics or audit stores.
Provider keys and AI routing live exclusively on the server. Deterministic clinical rules and the code-lookup indexes run locally in the extension — so checks are instant and work even when the server is unreachable.
Every suggestion, acceptance, and rejection is logged with attribution type and source authority for complete defensibility.
// Analytics event — no PHI stored
{
"eventId": "evt-00127",
"eventType": "suggestion.applied",
"sessionId": "sess-abc-123",
"findingId": "f-0042",
"severity": "high",
"findingType": "icd10_specificity",
// No patient names, IDs, or clinical data
"verificationStatus": "verified",
"attributionType": "deterministic",
"sourceAuthority": 1
}Start with the local demo, evaluate with your own data, then deploy where your compliance and infrastructure requirements demand.
Bring your own model API key. Run entirely on your machine with demo EHR scenarios and full local control. Zero cloud dependencies.
Available NowManaged governed server with all AI providers included. Cross-device settings sync, automatic updates, and no infrastructure to manage.
CloudRun entirely on your machine with 13 synthetic clinical scenarios. Evaluate functionality before committing — no account required.
Try It FreeWindows: run the installer, restart your browser, done — the extension installs itself silently in Chrome and Edge and stays updated automatically. Manual zip install remains available for Windows, macOS, and Linux.
Double-click AIMedicalEvaluator-Setup.exe, click through, fully close and
reopen Chrome or Edge. The extension appears enabled for every profile — no developer
mode, no chrome://extensions, no drag-and-drop, and updates arrive automatically.
Grab ai-medical-evaluator-v2.1.0.zip and unzip it to a permanent location you control (e.g. C:\Users\<you>\MedEval\).
Navigate to chrome://extensions/ and toggle Developer mode on (top-right corner). Then click Load unpacked.
Works the same way in Edge at edge://extensions/.
Select the unzipped folder. The MedEval icon appears in your toolbar — pin it for one-click access. Open any web-based EHR form and click the icon to scan.
First run: enter your facility server URL & credentials in Settings.
Whether you are a facility evaluating the platform, a health system planning a pilot, or a payer interested in provider-side integration, we would like to hear from you.
In demo mode, messages are stored locally. No data is sent externally.