Patient Literacy Assessment
AI estimates a patient's reading level from intake answers.
- 36%: US adults, low health literacy (struggle with medical text)
- 7th–8th grade: Average US reading level (national baseline)
- 10th grade: Typical patient handout level (above patient average)
- 5–8: Intake questions used (short response items)
Intake signal collection
AI reads response length, vocabulary, and phrasing.
Literacy scoring model
Responses map onto a basic-to-advanced scale.
A short vital-sign style intake takes under two minutes.
Confidence and uncertainty
Low-confidence estimates trigger a clarifying question.
Content Complexity Tiers
The same medical fact exists at three complexity levels.
- 3rd–5th: Basic tier reading level (grade equivalent)
- 6th–9th: Intermediate tier reading level (grade equivalent)
- 10th–14th: Advanced tier reading level (clinical language)
- 3: Content variants per topic (pre-authored tiers)
Tier authoring
Clinicians pre-write each tier once, up front.
Icon-heavy basic tier
Short sentences pair with simple, familiar icons.
Every tier preserves the exact same medical meaning.
Clinical advanced tier
Precise terminology, dosages, and mechanism included.
Literacy-Level Matching
AI selects the tier that fits the patient's score.
- <200ms: Matching latency (real-time selection)
- ±1 tier: Mismatch tolerance (before override)
- 20%: Topic-complexity weighting (adjusts final tier)
- Intermediate: Default fallback tier (used when uncertain)
Score-to-tier mapping
Literacy score converts directly into tier choice.
Topic difficulty adjustment
Technical topics nudge selection toward higher tiers.
Matching blends patient score with topic-specific difficulty.
Clinician override option
Care teams can manually force a tier.
Content Rendering
Matched-complexity content displays instantly to the patient.
- <1s: Render time (tier selection to screen)
- ~40%: Reading time saved (vs mismatched tier)
- High: Icon usage, basic tier (supports low literacy)
- +25%: Comprehension lift, matched tier (vs generic content)
Format stays consistent
Layout stays the same across every tier.
Plain-language substitution
Jargon swaps for everyday equivalent words.
Right-tier content cuts re-reading and confusion.
Visual support
Icons and diagrams reinforce simplified text.
Adaptive Refinement
AI adjusts tier from engagement and comprehension signals.
- 4: Signals monitored (dwell, scroll, clicks, quiz)
- 30s: Refinement check interval (continuous monitoring)
- Fast, confident reading: Tier upgrade trigger (engagement pattern)
- Re-reading, long pauses: Tier downgrade trigger (engagement pattern)
Engagement tracking
Scroll speed and dwell time reveal difficulty.
Comprehension quizzes
Short quiz answers confirm true understanding.
Refinement keeps content matched as understanding changes.
Continuous tier adjustment
Model nudges the tier up or down live.