← Back to portfolio
PhD Project · UX Research · Human–AI Interaction · LLM Agents · Digital Health

Automating health education at the point of care.

I designed an end-to-end human-centered AI framework that combines conversational agents, structured user modeling, LLM-based content generation, grounded prompting, and teach-back to make patient education more personalized, interactive, and scalable.

1

AI-Driven User Understanding

Designed conversational systems that dynamically collect user context, behavioral patterns, literacy needs, goals, and decision-making factors to support personalized digital experiences.

2

LLM-Powered Personalization Engine

Integrated large language models with structured user modeling and grounded content generation to automate scalable, personalized health communication and education workflows.

3

Adaptive Human–AI Interaction Design

Designed interactive AI experiences that guide users through complex information, assess comprehension in real time, and dynamically adapt communication based on user responses.

4

Scalable Digital Health Automation

Operationalized evidence-based communication and learning strategies into scalable AI-driven systems that can support patient engagement, education, and behavior change across healthcare environments.

The Problem

Patient education is often generic and designed for the general population, making it difficult for many patients, especially those with low health literacy, to understand and apply. This can lead to poor health outcomes, including higher hospitalization and mortality rates, while also increasing healthcare system costs.

Research insight

Patients do not only need more information. They need information that is relevant to their situation, delivered in a way they can process, and reinforced through interaction.

Design opportunity

Automate best practices in patient communication—agenda setting, tailoring, teach-back, and take-home materials—so high-quality education can be delivered consistently at scale.

My solution

Collaborative AI Interaction

Designed human-centered conversational experiences that adapt to user priorities, goals, and information needs in real time.

Dynamic User Intelligence

Built structured user modeling systems that capture behavioral, contextual, and decision-making factors to drive personalization.

LLM-Powered Personalization

Integrated large language models to generate adaptive, personalized content and recommendations at scale.

Adaptive Guidance & Engagement

Designed AI-driven interaction flows that guide users through complex information using interactive and cognitively supportive experiences.

Continuous Understanding & Feedback

Implemented real-time comprehension assessment and adaptive feedback mechanisms to improve understanding, engagement, and decision support.

AI-driven Solution Architecture

Human–AI Workflow Architecture

Designed end-to-end human–AI workflows integrating conversational data collection, user modeling, LLM personalization, adaptive guidance, and comprehension support.

LLM System Design

Developed grounded LLM generation pipelines that combined structured user attributes with validated clinical content to deliver personalized, context-aware, and clinically aligned educational experiences while improving output relevance, reliability, and clarity.

Conversational AI & Adaptive Interaction Design

Designed adaptive conversational AI interaction frameworks for contextual information gathering, personalized education delivery, real-time comprehension assessment, and dynamic clarification strategies to support effective human-centered communication and learning.

T-HEALS framework showing agenda setting, data collection, LLM prompting, tailored care plan generation, care-plan walkthrough, teach-back, and take-home reference

Skills highlighted

AI/LLM system designStructured user model, personalization pipeline
Prompt engineeringGrounded prompts, content constraints, health context
Content groundingValidated sources, safety, consistency
Human–AI interactionTrust, agency, comprehension, engagement
Conversational UXDialogue flows, agenda setting, teach-back loops
Healthcare UX researchHealth literacy, patient education, behavior change
Mixed-methods researchSurveys, interviews, pre/post measures, PEMAT
Experimental designControl vs intervention, feasibility, summative evaluation
Information architectureCare-plan structure, content hierarchy, readability
AI evaluationAccuracy, clarity, usefulness, reliability
Workflow automationClinical best practices at scale
Product strategyFrom research framework to scalable AI experience

Why this matters

Healthcare teams need scalable ways to make education more personal, understandable, and actionable. This project shows how AI can be designed as a supportive communication layer—not to replace clinicians, but to help deliver consistent, patient-centered education when time and resources are limited.

UX Research Digital Health LLMs Conversational Agents Health Literacy Patient Education
Back to portfolio Connect on LinkedIn