TrialClinIQ
Collapsed a multi-step clinical alert process into a single AI dashboard, enabling 1-click patient flagging and automated trial matching directly from EHR data.
Clinical trial recruitment fails quietly because inclusion/exclusion protocols are written for regulators, forcing clinicians into high cognitive strain and manual EHR cross-checking during patient visits.
Leveraging my Pharmacology background, I deconstructed FDA and ClinicalTrials.gov protocols and designed a 0→1 dual-portal system with structured eligibility blocks, an AI interrogation drawer, and plain-language patient summaries.
Delivered 60+ high-fidelity screens (80+ including states) across 16 core flows and 2 parallel dashboard systems, achieving 100% clinical governance sign-off from healthcare stakeholders.
Overview
A B2B clinical intelligence platform helping health clinics and community healthcare providers (FQHCs) use data to automate patient specialty referrals and clinical trial matching. I designed the 0→1 product experience, from trial discovery through to recruitment decisioning.
Problem worth solving
Clinical trial recruitment fails quietly. Sponsors lose months, sometimes years, not because patients do not exist, but because the systems for finding and qualifying them are broken.
“How do you design confidence into an AI-driven matching system when the stakes are a patient’s care and a sponsor’s timeline?”
What I was brought in to solve
Transform complex AI clinical matching into a transparent, human-governed, and fast UI experience across two distinct user groups: clinical recruiters/investigators who need rapid audit trails, and patient volunteers who need clear, plain-language eligibility checks.
How I approached it
Leveraging my background in Pharmacology, I approached research not just as a UX designer, but with an acute understanding of how clinical protocols and medical data operate in the real world.
1. Protocol Deconstruction & Desk Research
Before drawing a single frame, I analyzed dozens of active trial protocols from ClinicalTrials.gov and FDA regulatory submissions.
- Key Takeaway: Protocols rely on conditional logic matrices (e.g., HbA1c > 7.5% AND no history of renal impairment within 12 months). The AI interface needed to expose this underlying logic, rather than hiding it behind a single percentage match score.
- Volunteers Core Need: Clear, empathetic language, simplified eligibility checks, and control over personal medical data sharing.
- Recruiters / Investigator Core Need: Rapid screening tools, instant audit trails, and zero friction to interrogate or override AI matches.
The output
Like any complex 0-to-1 system, the best design decisions came from testing and eliminating flawed approaches.
1. Grounded in Domain First: Structuring Clinical Criteria
I mapped inclusion and exclusion criteria into structured visual data blocks rather than walls of text.
- Eligibility Pill Badges: Green (Matched) and Red (Unmatched).
- Instant Logic Highlighting: Clinicians can inspect any eligibility criterion to see the exact EHR source data.
2. Progressive Disclosure & The Interrogation Drawer
To resolve the "Black Box" issue, I engineered a layered information architecture that lets clinicians interrogate or override AI matches without losing their place in the patient queue.
3. Dual-Dashboard Architecture
A. The Clinical Recruiter Terminal
Designed for high data density, quick keyboard shortcuts, and fast batch decisioning across multiple trial sites.
- Candidate Data Grid & Table Views: Filter candidate lists by trial status, compatibility score, or trial location.
- One-Click Actions: Schedule Prescreening, Request Additional Documents, and Add Notes directly from the primary table view.
B. The Volunteer Trial Discovery Hub
Designed for accessibility (WCAG AA compliant), clear typography, and patient agency.
- Plain-Language Summaries: AI translates dense medical jargon into 6th-grade reading level explanations ("This study tests a new daily pill for type 2 diabetes").
- Transparent Data Sharing: Clear consent toggles showing exactly what medical metrics are shared with trial coordinators.
Impact
TrialClinIQ is currently in pre-launch. Impact is measured through design scope and stakeholder validation:
What I’d tell the next designer
This project sits at the intersection of healthcare systems, AI, and human-centered design. My design work focused on making complex systems usable under real-world constraints: when clinicians cannot see why an algorithm matched a patient, they will not trust the match.