· Interactive decision aid (HTML/SVG + audio) · Audience: “Marcus, 58” — cautious, pill-averse, prefers numbers he can see AI-assisted · Clinical review required · Provenance label (illustrative) © HealthcareGPS Inc - Copyright 2026

Sample Health Plan · Your heart, your choice

Should I take a statin? See 100 people like you.

A statin is a daily pill that lowers cholesterol. It can lower your chance of a heart attack or stroke. Let’s look at what that means for people like you.

2Over the next 10 years…

0
have a heart attack or stroke
0
avoid one because they took a statin
0
do not have one either way

3What about side effects?

1
gets muscle aches caused by the statin. These usually go away if you stop or switch.
<1
is told they have diabetes a bit sooner. This is more likely if blood sugar is already high.
Most people have no side effect from the statin.
Illustrative — not medical advice

4What matters most to you?

There is no wrong answer. Slide to show what feels more important to you right now.

5Take this to your doctor

My questions about statinsSample Health Plan

What is my real 10-year chance of a heart attack or stroke? How much would a statin lower my chance? What side effects should I watch for, and what do I do if I get one? Can changes in food, walking, or quitting smoking help instead of, or along with, a pill? If I start, how long would I take it? Can I stop later?

6Check: did we explain it well?

This is a check on us, not on you. If something was unclear, we explain it again.

Why this works

Icon arrays and natural frequencies. Showing risk as “12 out of 100 people” with pictures of people is easier to understand than percentages or relative risk. This is well-studied in risk communication (Zikmund-Fisher and colleagues; Garcia-Retamero and Galesic). We show both the “with” and “without” groups so the absolute difference is clear, not just a relative reduction.
Patient decision aid elements. It shows options, benefits and harms with equal detail, and helps people clarify their values, in line with International Patient Decision Aid Standards (IPDAS). Cochrane reviews (Stacey et al.) find decision aids improve knowledge and accuracy of risk perception.
Teach-back, built in. The check is framed as a test of our explanation, like the AHRQ teach-back method. Wrong answers get a plain re-explanation, then a second try.
Multimedia principles (Mayer). Words and pictures together; read-aloud is optional and the user controls it.

Behavior Change Techniques (BCT Taxonomy v1): 5.1 Information about health consequences 9.2 Pros and cons 1.2 Problem solving 4.1 Instruction on how to perform the behavior 3.1 Social support (unspecified)

COM-B target: Capability (psychological — understanding absolute risk) and Reflective motivation (values-based choice). The goal is an informed choice, not a forced “yes”.

Persona tuned for: “Marcus, 58” — high blood pressure, distrusts pills, likes to see the numbers himself, reads at about a 6th-grade level.

How HealthLiteracyCopilot would make it: HealthLiteracyCopilot would generate the copy and structure from a decision-aid template; the risk model is supplied by the organization’s clinical team (never invented by AI). It would then be checked with an adapted PEMAT actionability rubric for interactive content, the CDC Clear Communication Index (numbers and risk items) and an IPDAS checklist, and tried with simulated “Marcus” readers (simulated participants, not patients) to find questions for human review. The Spanish version gets qualified human review.