AI & Healthcare: Educator & Facilitator Guide
For health educators, patient advocates, social workers, community health workers, and legal aid staff. Facilitation guide for the Insurance Denial Navigator, discussion questions on algorithmic accountability in healthcare, and a patient advocacy simulation.
How to use the Denial Navigator in a group.
The tool is most powerful when run as a shared exercise — using a real or realistic denial scenario.
Setup (5 minutes)
- →Print or display a sample Explanation of Benefits (EOB) and denial letter — these are publicly available from CMS and several patient advocacy organizations. Ask participants to identify the denial type from the letter.
- →Ask for show of hands: 'Has anyone in this room, or their family, received an insurance denial?' In most rooms, this is 40–60% of participants.
- →Frame the exercise: 'We're going to walk through what you can actually do when you get a denial. Most people don't know they have appeal rights — and most appeals, when filed correctly, succeed.'
Discussion questions during navigation
- →"If an algorithm denied your claim in 0.4 seconds after reviewing your records for less time than it took you to read this sentence — is that a medical decision?" — Surfaces the difference between automated processing and clinical judgment.
- →"The insurer says your treatment isn't medically necessary. Who defined 'medically necessary' — your doctor, or the insurer's algorithm trained on cost data?" — The criteria question.
- →"Your appeal deadline is 30 days from the denial letter. The letter came on day 7 of the month. When did your 30 days start — the denial date, the postmark, or the date you received it?" — Makes the procedural complexity real.
- →"An external reviewer is independent of your insurer. They reversed 40–60% of denials. Why doesn't everyone file for external review?" — The awareness and complexity gap.
- →"If you're 91 years old and got discharged from rehabilitation care against your physician's recommendation, what would you do next?" — The Gene Lokken case made concrete.
Debrief (10 minutes)
- →Review the three most actionable items: request specific criteria in writing, file internally and externally simultaneously, ask whether AI was used.
- →Name the Patient Advocate Foundation explicitly: it is free, assigns a real person, and works complex denials most patients cannot navigate alone.
- →Connect to policy: 'California now bans fully automated denials for certain services. Texas requires disclosure. Federal law requires independent external review. These rights exist because advocates fought for them.'
- →Leave participants with one concrete next step they could take today — not 'if they get a denial someday' but an action relevant to where they are now.
75-minute workshop structure.
Tested format for community health centers, legal aid organizations, senior centers, and union health and safety committees.
Opening: What does an insurance denial look like?
Display a real or realistic denial letter. Ask participants to identify: who signed it (often no one), what reason was given, what it says you can do next. Most denial letters are designed to discourage appeal.
The AI problem
Gene Lokken case. How nH Predict works. The 90% denial rate. The 13% improper denial finding from OIG. Frame: this is not isolated — it is systemic design.
Navigation exercise
Run the Insurance Denial Navigator as a group for two scenarios: an AI-generated prior authorization denial, and a medical necessity denial for an urgent situation. Pause at each step for discussion.
Racial and economic disparities
Obermeyer et al. (Science 2019) — the commercial algorithm that underidentified Black patients as high-need. Diagnostic AI performance gaps. The structural feedback loop.
Policy landscape and advocacy
California SB 1120, PARAM Act, CMS enforcement of Medicare Advantage rules. Map the levers: federal regulation, state law, litigation, procurement standards.
Action commitments
Each participant leaves with: the Patient Advocate Foundation URL saved, the CMS external review page bookmarked, and one concrete action — file a records request, contact a state legislator, or share the navigator with one person they know who has a pending denial.
Adjusting for your audience.
Senior center / Medicare Advantage audience
- →Focus on: Medicare Advantage specifically — OIG findings, the 13% improper denial rate, and prior authorization reform.
- →Connect to: the AARP campaign on Medicare Advantage prior authorization reform.
- →This audience has direct stakes — many are in MA plans. Make the appeal process immediately actionable.
Community health center / uninsured patients
- →Focus on: Medicaid managed care AI denials and the parallel process. The appeal rights are similar but the resources differ.
- →Name specific local resources: legal aid, patient navigator programs, and community health worker referrals.
- →ACA marketplace plans have the same external review rights — but Medicaid managed care varies by state.
Healthcare workers and clinicians
- →Focus on: clinical decision support AI embedded in EHR systems they already use. Sepsis prediction, readmission risk scores, medication alerts.
- →The Obermeyer et al. paper is most relevant here — the algorithm they used is in clinical settings nationwide.
- →Discussion: when should clinicians override an AI recommendation, and how is that decision documented?
Policy and legal audience
- →Focus on: ERISA §503, ACA §2719 enforcement gaps, and the PARAM Act.
- →The litigation landscape (UHC class actions) and what plaintiffs are arguing about the AI-denial feedback loop.
- →Congressional testimony from the 2023 Senate investigation is publicly available and useful primary source material.
Educator references.
Want CPAI to deliver this workshop in your community?
We partner with community health centers, legal aid organizations, senior centers, and labor unions.