Skip to main content
Scienza Health
Clinical AI

V28 and Dementia: In an Encounter or Not at All

David Kaiser, Founder & CEO··Updated October 2026·4 min read
Share

Under the V28 risk adjustment model, a Medicare Advantage member's dementia counts when a clinician documents it in a clinical encounter. Finding the members who need that encounter is the part nobody has staffed.

Last updated: October 2026

Under the V28 risk adjustment model, a Medicare Advantage member's dementia counts when a clinician documents it in a clinical encounter. Finding the members who need that encounter is the part nobody has staffed.

Why does dementia have to be documented in a clinical encounter under V28?

V28 is fully phased in for payment year 2026, and under it a member's dementia is recognized when it is documented in a face-to-face or qualifying clinical encounter, not when it appears in a chart review or a care manager's note. A plan that wants its members' cognitive status reflected accurately has to get each of those members in front of a clinician who assesses them and writes it down.

The old habit was to find conditions in the chart after the fact. V28 closes that door for dementia. The condition is recognized through an encounter, which means a clinician, a member and a documented assessment.

V28 also splits dementia into three severity groups: HCC 125 for severe, HCC 126 for moderate and HCC 127 for mild or unspecified. Severity is a clinical judgment, which is one more reason the assessment belongs in an encounter rather than a retrospective review.

Nothing in this post is about a plan's risk score or payment. Scienza does not make claims about either. The question here is operational: how a plan gets members with unrecognized cognitive decline into a clinical encounter.

That question has two halves. Finding the members is an analytics problem the plan's data can answer. Reaching them is a staffing problem the plan's data cannot. Most plans have invested in the half they can see on a dashboard and left the other half to care managers who are already over capacity.

Why the members who need the encounter are the hardest to find

A member with early cognitive change does not call to report it. She misses an appointment, then another. She answers the annual survey politely. Her daughter notices, and the plan never hears from the daughter.

Care managers are working a list that was built from claims. Claims show what has already been coded. The member who has not yet been assessed is, by definition, not on that list.

The history that precedes a dementia transition is in the plan's data anyway: coded conditions, encounters, medications, assessments. Scienza's foundation model for aging reads it and returns an aggregate one-year dementia report for the cohort. Members are never named. It points care management at a cohort. A clinician does the rest.

What a between-visit check-in adds

A report says where to look. It does not reach anyone. Gia does: by phone or video, any phone including a landline, in 44 languages, with retries for missed calls.

Inside the conversation runs a 40-second speech check-in that measures 2,500 speech biomarkers: pace, pauses, articulation and prosody. How a member sounds today is information the record does not hold. Results reach the care manager in under two minutes, as a ranked worklist with the reason beside each name.

If the clinician wants more before scheduling, one click on Initiate Assessment sends Gia back to administer a structured cognitive assessment in conversation. The clinician asked a question and received an answer, without a visit.

Caregiver check-ins run alongside. The daughter who has quietly taken over the pill box gets a call too, and what she has noticed sits beside what the member reported and how the member sounded. For a plan trying to decide who comes in this month, that is three sources where the chart had none.

The encounter happens with your clinician

Gia does not document a condition and does not perform the plan's health risk assessment. Gia surfaces who to assess and brings the clinician what she needs before the visit. The encounter, the assessment, the severity judgment and the documentation are the clinician's.

Each check-in result arrives with Proactive Decision Actions. Options, never orders: schedule a visit, send a nurse, request a medication review, or wait. The clinician's choice writes back to the EHR, dated and attributed. Gia is clinical decision support, not a diagnosis.

For D-SNP members, the same call can carry the outreach the plan owes anyway. Under 42 CFR 422.101, integrated D-SNPs owe every member an annual comprehensive health risk assessment and an individualized care plan. Gia does not perform the HRA. Gia gets the member on the phone, which is the step staff cannot get to.

What the record shows at year end

Every member reached, every attempt that went unanswered, every clinician decision and every encounter that followed, dated and attributed. When a reviewer asks why a member's cognitive status was recognized this year and not last, the AI co-clinician record shows the check-in, the worklist and the clinician who acted on it.

Where to start

Choose members over 75 with no cognitive assessment in the record. Run the dementia report on your own data, have Gia check in with the members your team chooses, and bring the ranked worklist to care management with the encounter calendar open.

See how Medicare Advantage and D-SNP plans use Gia, then book a working session.

Sources & References

  1. CMS, Medicare Advantage risk adjustment: https://www.cms.gov/medicare/payment/medicare-advantage-rates-statistics/risk-adjustment
  2. eCFR, 42 CFR 422.101 (health risk assessments and care plans for dual eligible special needs plans): https://www.ecfr.gov/current/title-42/chapter-IV/subchapter-B/part-422/subpart-C/section-422.101
DK
David KaiserFounder & CEO

David Kaiser is the Founder and CEO of Scienza Health, the company behind Gia, an AI co-clinician that health plans, ACOs, health systems and senior-care operators use to check in with older adults between visits.

This content is intended for informational purposes and does not constitute medical advice. Citing research does not mean this page or the product has been independently peer reviewed.

Share

See Gia work on your population.

Predictive medicine, from home care to health systems.

Book a demo