To build an ACO quality report with AI, export your attribution file, claims or visit data, and gap reports. Upload them to a HIPAA-compliant AI workspace. Ask for each metric in plain language: attribution counts, AWV completion, gap closure by measure, adherence and TCM timeliness. Check the math against a few known rows, then export to Excel or Power BI.
Takeaways
Use an AI tool with a BAA
Your quality report runs on patient data. Consumer chat tools without a business associate agreement aren't an option.
Six sections cover most committees
Attribution, AWVs, gap closure, adherence, TCM timeliness and high-risk patients without a visit.
Check the numbers before anyone sees them
Spot-check counts against the source file. AI can misread a column.
The report should end in a call list
Every gap row is a patient someone needs to reach.
What goes in a monthly ACO quality report
An accountable care organization (ACO) quality committee needs the same core picture every month: who you're responsible for, what's open, and whether it's getting better. These six sections cover it.
1. Attribution
Attributed patients this month vs last month. Adds and drops by practice and provider.
2. AWV completion
Share of attributed patients with an annual wellness visit in the last 12 months. By practice and provider.
3. Gap closure by measure
Open and closed gaps for each quality measure. Blood pressure control, diabetes control, cancer screenings, depression screening.
4. Medication adherence
Patients below 80% proportion of days covered (PDC) for diabetes, blood pressure and statin meds.
5. TCM timeliness
Discharges with contact inside 2 business days. Face-to-face visits inside 7 or 14 days.
6. High-risk, no visit
Patients with high risk scores or multiple chronic conditions and no visit this year.
Not every committee needs all six. Start with what your contract pays on. See our attribution glossary entry if your team is new to the term.
What MSSP measures on quality
If you're in the Medicare Shared Savings Program (MSSP), your quality score runs through the APM Performance Pathway (APP). APM means alternative payment model. The measure set is called APP Plus. Read our MSSP glossary entry for background.
Two changes from the 2026 Quality Payment Program rule matter for your report:
- SDOH screening removed. CMS removed Screening for Social Drivers of Health (Quality ID 487) from the APP Plus set. You can drop it from your committee dashboard.
- Medicare CQM population narrowed. Medicare CQMs are clinical quality measures. A patient now counts as eligible only with at least 1 primary care service in the performance year. That applies from performance year 2025 on.
The second change has a direct operational effect. An attributed patient with no primary care visit this year isn't in your Medicare CQM denominator. Section 6 of your report is a quality list too, not only a cost list.
The APP Plus set also includes diabetes glycemic status, depression screening, blood pressure control, cancer screenings, readmissions and the CAHPS patient survey. Check CMS's current specifications before you map local metrics to them.
Building the report, step by step
Step 1: Pull the exports
You likely already have the files. Typical sources:
- Attribution or assignment file from CMS, your ACO, or the health plan
- Claims or encounter export from your EHR or billing system, with visit dates and CPT/HCPCS codes
- Gap report from the plan or ACO, listing open measures by patient
- Pharmacy adherence report from the plan, with PDC by measure
- Discharge or ADT log with discharge dates and first-contact dates
Export each as CSV or Excel. Keep the original column headers. Don't clean them up by hand first. The AI can map messy headers, and you'll want the originals to check against.
Step 2: Use a workspace you're allowed to put PHI in
This is the step teams skip. OpenAI says consumer ChatGPT plans (Free, Plus, Pro) and ChatGPT Business aren't eligible for its BAA. Consumer Gemini isn't covered either; Google warns that human reviewers may read chats. Pasting patient rows into those tools sends protected health information (PHI) to a vendor with no business associate agreement (BAA). That's a HIPAA problem regardless of how good the output is. See our comparison of whether ChatGPT is HIPAA compliant.
Rivvi's HIPAA-compliant AI workspace includes a BAA on the free plan, for you and two colleagues. It's SOC 2 Type II. Patient data isn't used to train third-party models, and PHI access is logged. Uploads are enriched against CMS and NPI registry data.
Rivvi also keeps memory of your organization. Your measure definitions, practice names and report format carry over month to month. You can view, edit or delete what it remembers.
Step 3: Ask for each section
Write prompts the way you'd brief an analyst. Be specific about the time window and the denominator.
- 1
Attribution
Using the September and October attribution files, count attributed patients by practice and provider. List patients added and dropped since last month.
- 2
AWV completion
For each attributed patient, find the most recent G0438 or G0439 in the claims export. Show the percent with an AWV in the last 12 months, by provider.
- 3
Gap closure
From the gap report, count open and closed gaps for each measure. Compare to last month's report. Flag measures that got worse.
- 4
Adherence
From the pharmacy report, list patients under 80% PDC for diabetes, RAS antagonist and statin measures. Sort by days until the end of the year.
- 5
TCM timeliness
For each discharge in the log, calculate business days to first contact and days to the follow-up visit. Show the percent within 2 business days, 7 days and 14 days.
- 6
High-risk, no visit
List attributed patients with 2 or more chronic conditions and no primary care visit in 2026. Include phone number and assigned provider.
Save prompts that work. Next month, ask Rivvi to run the same report on the new files.
Step 4: Check the numbers
AI tools can misread a column, double-count a patient, or apply the wrong date range. Before the report goes to committee, run these checks:
- Totals match the source. Attributed patient count should equal the row count of the attribution file after removing duplicates.
- Spot-check 5 patients. Pick names at random. Confirm their AWV date, gap status and visit history by hand.
- Ask how it calculated. Ask Rivvi to show the formula and the columns it used. If it used the wrong date column, correct it. Rivvi's memory can keep that correction for next month, and you can view or edit what it holds.
- Watch the denominators. For percentages, confirm who's in the bottom number. "Of all attributed" and "of those eligible" give different answers.
- Compare to last month. A jump of 20 points in one month usually means a data problem, not a performance win.
Step 5: Build and share the report
Ask Rivvi to assemble the sections into one report. Put a short summary at the top: what improved, what got worse, and the 3 things to act on. Then share it where your team works.
- Excel: export the tables for the committee packet.
- Power BI: push to Power BI if your organization uses it for dashboards.
- SharePoint or OneDrive: save the report alongside last month's.
- Microsoft Teams: staff can tag @Rivvi in a channel to ask follow-up questions.
Rivvi connects to Microsoft (Power BI, SharePoint, OneDrive, Excel, Outlook) and Google (Drive, Sheets). Email is draft-only. Rivvi drafts the cover note and a person sends it. See all integrations.
Turn the report into action
A quality report that doesn't change who gets called next week is just a record. Sections 3, 4, 5 and 6 each produce a list of patients. On the Team plan, Rivvi calls and texts those patients. It warm-transfers them to your scheduler or care team, or creates a follow-up task, and logs each outcome. Next month's report then shows whether the outreach worked. For the list-building side, see our guide to building a care gap list from an attribution file.
Build this month's quality report today
Upload your attribution, claims and gap files and ask for the report. Free for you and two colleagues, with a HIPAA BAA.
Sources
- CMS: 2026 Quality Payment Program Final Rule Fact Sheet and Policy Comparison Table
- CMS MLN908628: Transitional Care Management Services (August 2025)
- eCFR: 42 CFR 410.15, Annual wellness visits
- PQA: Measures Resources
- OpenAI: HIPAA Implementation and Configuration Guide
- Google: Gemini Apps Privacy Hub