Case Study 03 NIH-Funded Research Health Equity Systems Design Longitudinal Study

Field Guide

Designed the digital field data collection system for a 700-household NIH-funded cohort study, turning fragile paper infrastructure into the operational backbone for a landmark study in dental disease genetics.

200%
Enrollment increase
82%
Faster completion
0%
Data degradation
Context
01 · Background

High-impact science.
Fragile infrastructure.

This was a multi-site NIH/NIDCR-funded longitudinal study examining health disparities across rural Appalachia, collecting data across 700 households with children ages 1–18.

Role
Lead Experience Designer
Partners
University of Pittsburgh · West Virginia University
Platform
Voice-enabled tablet
Offline-capable · Bluetooth sync to central research database
Domain
Oral health disparities
Intergenerational stress · Genetic research · Health equity
Problem
02 · Discovery

The team said literacy.
The data said overload.

Paper surveys and manual workflows were creating random answer patterns beginning at the 34% completion mark, hours-long sessions, and families leaving before receiving the care they were promised. The research team assumed the issue was reading ability. That wasn't the full picture.

What they assumed
A literacy problem
Participants couldn't understand the questions
Rural populations were disengaged from research
Data degradation was random and unavoidable
Scheduling delays were an operational fact
What was actually happening
A systems problem
Postgraduate-level surveys created cognitive overload, not confusion
The 34% degradation pattern was statistically consistent and a signature of overload
20+ minute manual intake compressed clinic time and broke promises
Appalachian communities had warranted skepticism toward outside researchers - trust had to be earned operationally
Solution
03 · Design

A trusted guide,
not a digital form.

We designed a voice-enabled, tablet-based Field Guide that transformed dense research protocol into a structured, human-centered workflow. The core design principle: every system decision must either build trust or keep a promise. If it did neither, we cut it.

1
Preloaded Family Profiles
Call-center staff captured household structure during scheduling, eliminating the 20+ minute intake bottleneck at the point of care.
Reduced protocol friction before participants even arrived
2
3rd-Grade Reading Level Rewrite
Every question rewritten for clarity while preserving NIH scientific integrity. Eliminated cognitive overload without changing what was being measured.
Halted the 34% random-answer degradation pattern entirely
3
Voice + Headphones
Speech-enabled interface allowed participants to listen instead of read. Private audio playback maintained HIPAA compliance in shared clinic environments.
Increased accessibility across low-fluency populations
4
Offline-Resilient Data Capture
Tablet-based system supported structured data collection in low-connectivity rural environments with Bluetooth syncing to clinic systems.
Preserved data integrity across all rural sites
5
Smart Workflow Transitions
System-guided participant movement between survey and exam spaces ensured families received the care they were promised — the broken promise that was quietly destroying trust.
Reduced trust breakdown tied to scheduling delays
Architecture
04 · System Design

From deterministic capture
to agentic orchestration.

The original system replaced paper with a structured tablet workflow. A reimagined v2 explores how longitudinal research infrastructure evolves when agents can detect fatigue, adapt phrasing, reconcile contradictions, and coordinate multi-actor flow in real time.

V1 · Deterministic System

V1 Deterministic System — field setting with participants, researcher, and examiner feeding into a tablet UI with survey engine, validation, and offline store, syncing via Bluetooth to a central research database with bio samples and lab data Field Setting Participants Researcher Examiner Tablet System Tablet UI Voice · Touch Survey 120+ items Validation Required fields Offline Store Encrypted cache · Sync queue Data Platform Bluetooth Batch sync Research DB Central store Bio Samples DNA · Microbial Lab Data Fluoride · Env.
V1 — Deterministic capture: tablet UI → survey engine → validation → offline store → sync → research database

V2 · Agentic Evolution

V2 Agentic Evolution — field actors feed into a multi-agent layer including conversational capture, trust and engagement, data quality, and clinic flow agents, coordinated by a research orchestrator with longitudinal memory, outputting to research database, insight engine, audit layer, and human-in-the-loop review Field Participants Researcher Examiner Clinic Staff Agent Layer Conversational Capture agent Trust & Engagement Fatigue detection Data Quality Anomaly · Contradiction Clinic Flow Operational coordination Research Orchestrator Policy enforcement · Routing Longitudinal Memory · Structured DB · Vector store · Multi-year state Outputs & Oversight Research DB Central store Insight Engine Risk flags · Patterns Audit Layer Compliance trail Human in Loop Review · Override
V2 — Agentic evolution: multi-agent orchestration with fatigue detection, contradiction flagging, HITL review, and longitudinal memory
Impact
05 · Results

Field Guide didn't just improve UX —
it stabilized research infrastructure.

2×
Participation rate
Families returned year after year. Longitudinal cohort sustained across multiple sites.
82%
Faster completion
60+ minute sessions reduced to under 12 minutes on average.
0%
Degradation pattern
The 34% random-answer signature — a hallmark of overload — was eliminated entirely.
Same
Day data availability
Replaced multi-day paper processing. Data was analysis-ready by end of each clinic day.
NIH
Continued funding
The study secured continued NIH funding. Data integrity was no longer a risk to the research program.
GWAS
First in dental disease
Clean data enabled the first genome-wide association study of any dental disease.
Reflection
06 · What I learned

Three principles that inform
every system I build.

01
Question the frame
The team said "literacy." The data said "overload." The system said "broken promises." The real problem is rarely the stated problem — and the reframe is where the design work begins.
02
Design for dignity first
Trust compounds, or collapses, through operational follow-through. Every missed promise creates compounding skepticism. Every kept promise opens the door wider.
03
Technology serves relationship
The best systems feel like a guide, not a form. That philosophy — earned here in rural Appalachia — now informs every AI system I build, from coaching agents to enterprise decision tools.