← Back to portfolio
PhD Project · Mobile health · Virtual agent evaluation

Atrial Fibrillation Virtual Agent

A smartphone-based virtual coach designed to support people with atrial fibrillation through long-term education, medication adherence, symptom reporting, and heart rhythm monitoring

1

Human-AI Patient Engagement

Designed and evaluated conversational health experiences that supported ongoing patient education, self-management, and longitudinal engagement.

2

Behavioral Data Infrastructure

Worked with large-scale interaction, behavioral, and monitoring data including engagement patterns, reporting behaviors, and longitudinal system-use metrics.

3

SQL & Longitudinal Analytics

Applied SQL and quantitative longitudinal analysis to evaluate sustained engagement, feature utilization, retention, and user behavior over time.

4

Digital Health Product Insights

Translated real-world behavioral data into UX and product insights for conversational AI, remote monitoring, and chronic disease management systems.

The problem

Atrial fibrillation is a complex chronic heart condition that requires ongoing self-management. Patients need support for understanding the condition, taking medications, recognizing symptoms, and regularly using heart rhythm monitoring tools. Many mobile health systems struggle with sustained engagement, especially beyond the first few weeks.

Long-term engagement is hard

Voluntary-use health apps often decline in use over time, but chronic disease management requires repeated use and ongoing support.

Monitoring behavior matters

For AF management, regular ECG readings can help patients and clinicians understand heart rhythm events and symptoms.

Data alone is not enough

The design challenge was not only collecting logs, but interpreting what those logs reveal about engagement, adherence, and patient experience.

The solution

The AF virtual coach combined a smartphone-based embodied conversational agent, a portable ECG sensor, a central database, and a clinician monitoring workstation. The agent provided education, encouraged ECG use, supported medication and symptom reporting, and used social conversation and storytelling to promote sustained engagement.

Patient-facing experience

  • AF education and self-care guidance
  • Medication adherence reporting through agent dialogue or app widgets
  • Symptom reporting and severity logging
  • ECG reminders and goal-setting support
  • Daily storytelling and social conversation to maintain engagement

Research-facing data

  • Session counts and timestamps
  • Dialogue topics and story use
  • Medication reports and reporting modality
  • Symptom logs and patient-reported events
  • Goal-setting behavior and ECG reading frequency
  • Self-report ratings of satisfaction, trust, liking, and perceived knowledge

Research process

Behavioral & Clinical Context Modeling

Mapped healthcare workflows, patient behaviors, and engagement objectives to understand how digital interventions support long-term condition management.

Data Engineering & SQL Analysis

Worked with large-scale interaction logs, behavioral data, and monitoring records using SQL and structured analytics workflows.

Longitudinal Engagement Analytics

Analyzed retention, engagement trends, feature adoption, reporting behaviors, and sustained system use across extended user interactions.

UX & Product Insight Generation

Translated behavioral and quantitative findings into design insights for conversational AI, patient engagement, and digital health product strategy.

Skills

Longitudinal User Engagement Analysis

Used SQL and behavioral analytics to evaluate sustained user engagement, interaction frequency, retention patterns, and feature adoption over time.

SQL

Healthcare Data & Behavioral Insights

Analyzed interaction logs, symptom reporting, medication adherence, and ECG-monitoring behaviors to understand how users engage with chronic disease management technologies.

Monitoring

Quantitative UX Research

Integrated longitudinal usage metrics, engagement trends, and patient-reported outcomes to evaluate trust, satisfaction, adherence, and system effectiveness.

SatisfactionEffectiveness

Health AI & Digital Health Relevance

Generated insights relevant to conversational AI, remote patient monitoring, adaptive engagement systems, and AI-driven chronic disease management platforms.

Conversational AI