Grandpa's new diabetes diagnosis

Evidence basis

• Listening removes the reading barrier. Spoken content lets people who struggle with print take in the message, and it fits into daily routines (cooking, driving, waiting rooms). The full transcript keeps it accessible for people who are Deaf or hard of hearing.
• Narrative and modeling. A family conversation is a form of entertainment-education: the listener sees someone like them asking the same questions and making a plan (social cognitive theory: vicarious learning and self-efficacy).
• Teach-back built in. Tomás restates the three key points in his own words at the end, which models the AHRQ teach-back method for the listener.
• Cultural and language fit. The Spanish version is written for meaning, not word-for-word, with familiar foods (tortillas, rice, beans) on the plate.

Behavior Change Techniques (BCT Taxonomy v1)

BCT	Where
5.1 Information about health consequences	"Too much sugar … can hurt your eyes, kidneys, feet and heart."
4.1 Instruction on how to perform the behavior	Plate method; 10-minute walk after dinner
8.7 Graded tasks	Start with 10 minutes, build to 30
3.2 Social support (practical)	"I'll walk with you on Tuesdays and Thursdays."
1.4 Action planning	What to bring and ask at the first visit
6.2 Social comparison / 13.1 Identification of self as role model	Tomás shares his own plan

COM-B target

Psychological capability (what diabetes and A1c mean), social opportunity (family support), and reflective motivation (fear reframed into "I can do a lot").

How HealthLiteracyCopilot would generate, personalize and assess it

• Generate: a two-host script from approved clinical source content, written at a grade 5–6 level in each language, then voiced with disclosed synthetic voices (or recorded by real community voices from the same script).
• Personalize: the author picks the language, food culture and caregiver relationship for the audience, and each version is reviewed. The voice-note version is short and has no personal health information in it.
• Assess: PEMAT-A/V on the audio + transcript, CDC Clear Communication Index on the script, speech rate checks (about 140–160 words per minute), and simulated older Spanish-preferred listeners (simulated participants, not patients) to find confusing terms (for example, "A1c") for human review.

What a pilot would measure

Proximal: listen-through rate, a 3-question recall check, and confidence to manage diabetes. Distal: diabetes self-management education (DSMES) enrollment, first-visit attendance, A1c test completed, and eye exam completed (quality measures). Measures a pilot would track, not results or promised effects.