Use Case 1
Daily Pre-Visit Brief Generation
Trigger: Scheduled patients on tomorrow's provider panels
1.Pull FHIR bundle for each scheduled patient (conditions, observations, medications, encounters, document references)
2.De-identify clinical notes under HIPAA Safe Harbor
3.Run NLP extraction with confidence scoring (0.0–1.0)
4.Auto-map concepts ≥0.90 to ICD-10 and V28 HCC crosswalk
5.Queue 0.70–0.90 concepts for certified coder review with source evidence
6.Validate MEAT criteria using structured data and notes
7.Assemble advisory brief: confirmed opportunities, suspected items, documentation instructions
8.Route to Medical Director for clinical review before EHR worklist delivery
✓
Every physician opens their panel with a validated, evidence-backed HCC brief ready to guide documentation during the encounter.
Use Case 2
Post-Encounter Audit
Trigger: Physician signs SOAP note after patient visit
1.Auto-generate coded claims draft from structured EHR diagnoses
2.Compare draft against SOAP note for discrepancies
3.Flag codes lacking MEAT evidence in the note
4.Flag codes documented in note but missing from claims
5.Queue flagged items for certified coder review within 5 business days
6.Write complete audit trail with reference numbers for CMS compliance
✓
Zero undocumented submissions and a defensible audit trail on every claim.
Use Case 3
Portfolio RAF Migration Tracking
Trigger: Nightly aggregation of finalized HCC submissions
1.Pull all finalized HCC submissions from audit module
2.Apply both V24 and V28 RAF weight tables
3.Identify high-impact code transitions (diabetes splits, CKD granularity, CHF sub-typing)
4.Calculate revenue at risk from uncaptured codes
5.Detect systemic documentation gaps by provider
6.Generate executive dashboard for VBC leadership
✓
Leadership sees V24→V28 migration progress, revenue impact, and provider-level documentation patterns every morning.
Integrations
EHR / FHIR R4 API (Epic, Cerner, athenahealth)
CMS V24 / V28 HCC Crosswalk Tables
Clinical NLP Engine
Certified Coder Review Queue
Audit & Compliance Module
BI / Executive Dashboard
Industries
healthcare
See It In Action
A sample conversation with this agent
Prepare the pre-visit HCC briefs for Dr. Patel's panel tomorrow.
Pulling FHIR bundles for Dr. Patel's 14 scheduled patients. For each I'm extracting diagnoses, labs, vitals, meds, and notes across a 5-month lookback, then running NLP against de-identified notes to surface suspected HCCs. I'll map each concept to ICD-10, cross-reference against the CMS V28 HCC crosswalk, and validate MEAT criteria. You'll get 14 structured briefs showing confirmed opportunities, suspected items needing physician judgment, and specific documentation instructions. Medical Director approval is required before I route briefs to the physician's EHR worklist.
Why are you suggesting diabetic nephropathy for Martha Chen? She's just coded as diabetes in our system.
Her chart shows: active metformin and empagliflozin, two eGFR readings in the last 90 days (58 and 54), urine albumin-to-creatinine of 68, and a progress note mentioning 'microalbuminuria monitoring.' That's Monitored + Evaluated + Treated evidence supporting E11.22 (Type 2 diabetes with diabetic chronic kidney disease). My confidence score is 0.87, which puts it in the 0.70–0.90 coder review band—so this is a suggestion for physician judgment, not a diagnosis. The specificity shift from E11.9 to E11.22 carries a meaningful RAF weight delta under V28.
What's the portfolio revenue impact if we close all suspected items this quarter?
Across 4,218 attributed members, I'm tracking 1,847 suspected HCCs at ≥0.70 confidence. If validated and documented, projected V28 RAF delta is +0.184, translating to approximately $1.9M in incremental annual revenue at current CMS benchmark rates. The highest-impact transitions are diabetes specificity splits (487 patients) and CHF sub-typing (203 patients). I'll flag the documentation gaps by provider in tonight's portfolio dashboard.