Skip to content
caesarhealth
← Blog

Policy and primary care

AI-Enabled Annual Wellness Visits: Closing the Loop from Risk Detection to Delivered Care

Caesar Health · October 14, 2026

A summary of the white paper PCP 1st, LLC and Caesar Health, Inc. jointly submitted to the CMS CCSQ QualTech 2026 showcase on September 3, 2026, under Opportunity Area #2: Leveraging AI to Increase Annual Wellness Visits in Medicare.

The Medicare Annual Wellness Visit (AWV) is good at finding risk. Its Health Risk Assessment surfaces falls, memory concerns, hearing loss and functional decline. What happens next is the weak point. In one academic health system, among AWVs where patients newly reported memory concerns, a memory screener was documented in 4.7% of visits and a specialist referral in 1.3% (Wec et al., Innovation in Aging, 2025).

The paper's argument fits in one line: the unit of work is not the visit. It is the path from a detected risk to delivered care.

The four functions

  1. Find and reach. Eligible beneficiaries are identified from the practice panel and checked against CMS's HIPAA Eligibility Transaction System (HETS) before any outreach, so the practice knows the eligibility date and whether the visit is initial or subsequent. SMS and voice outreach, with telephone and staff-assisted options, completes the Health Risk Assessment before the visit.
  2. Prepare the encounter. Caesar Health assembles structured EHR data, notes, labs, medications and remote monitoring data into a prefilled AWV summary and a draft personalized prevention plan for the clinician to review.
  3. Close the loop. Every positive finding becomes a tracked follow-up action, an assessment, order or referral, drafted for clinician approval and routed into PCP 1st's chronic care management, remote patient monitoring, CGM and wearable workflows. Each finding is tracked until it is completed, declined or clinically closed. Open findings resurface on a set cadence.
  4. Code it correctly. The completed visit is billed as G0438 (initial) or G0439 (subsequent) before claim submission. CMS reported a 24.5% overpayment rate for AWVs in the 2024 reporting period, about $307.5 million in projected overpayments (CMS).

The clinician stays in charge

AI drafts. Clinicians validate findings and approve every patient-specific recommendation. The system does not diagnose, prescribe, refer or finalize an AWV on its own. Every AI-generated statement is traceable to its source, and every action is logged.

What to measure

In priority order:

  1. Loop-closure rate: the share of positive Health Risk Assessment findings that end in a completed follow-up action, by risk domain.
  2. Time from a positive finding to completed follow-up.
  3. AWV completion, reported separately for initial and subsequent visits.
  4. Health Risk Assessment completion before the visit.
  5. Clinician and staff preparation time per AWV.
  6. Medication reconciliation completion and preventive gaps closed.

Who it is for

Organizations that already carry Medicare populations: accountable care organizations, Medicare Advantage plans, primary care practices and clinically integrated networks. The paper proposes a phased start with one Medicare population and one EHR environment. Eligibility verification, outreach and AWV prefill come first. Follow-through tracking is layered in as findings accumulate.

The same follow-through layer runs the high-risk, high-cost patient program for ACOs that PCP 1st and Caesar Health launched in October 2026.

Related

This is a summary. The detailed architecture in the submitted paper is proprietary to Caesar Health, Inc. and PCP 1st, LLC. Submission to QualTech does not imply CMS endorsement.

Keep reading