Overview

Care Intellect is an AI-powered oncology platform designed to support clinicians throughout the patient journey. The broader product enabled oncologists and care coordinators to review patient summaries, treatment plans, and treatment adherence.

Before any of that information became actionable, clinicians first needed to determine whether a patient was eligible for a treatment or clinical trial. I owned the design of this Clinical Trial Eligibility experience, focusing on creating an AI-assisted workflow that clinicians could trust.

The Problem

Evaluating clinical trial eligibility is one of the most time-consuming tasks in oncology.

Clinicians manually compare patient records, laboratory reports, pathology results, and imaging data against lengthy eligibility criteria. The process often takes weeks and is prone to inconsistency.

While AI had the potential to automate much of this work, a single AI-generated decision wasn't enough. In healthcare, clinicians need to understand why a recommendation is made before they can trust it.

Key challenges:

Eligibility criteria are complex and highly conditional
Patient data exists across multiple clinical sources
Clinicians needed transparency—not just automation
AI recommendations had to support, not replace, clinical judgment

Research

Instead of immediately designing detailed screens, our team decided to validate different approaches through exploratory research.

Exploratory Concept Evaluation

We created multiple design concepts representing different ways clinicians could evaluate eligibility.

These concepts were evaluated with oncologists and care coordinators to understand how they naturally reasoned through clinical decisions before committing to a final interaction model.

Cross-functional Collaboration

Throughout the project, I worked closely with researchers, usability specialists, AI engineers, and product stakeholders to ensure the experience balanced user needs with technical feasibility and the evolving capabilities of the AI models.

Concept exploration with clinicians Placeholder — swap in concept sketches or a research session photo.

Key Insights

Research revealed that clinicians weren't asking AI to make decisions for them.

They wanted AI to reduce the effort of gathering and organizing evidence while keeping them in control of the final decision.

Three insights shaped the experience:

Transparency builds confidence more than automation.
Clinicians think criterion by criterion—not as a single pass/fail decision.
Supporting evidence is just as important as the recommendation itself.

Design Process

Defining the Experience

Based on our research, I shifted the experience away from presenting a single eligibility verdict.

Instead, AI evaluated each eligibility criterion individually, allowing clinicians to quickly understand:

Which criteria were satisfied
Which required additional information
Which evidence supported the AI's recommendation
Where manual review was still required

Iteration & Validation

As the AI capabilities evolved, the design evolved alongside them.

We iterated on interaction patterns, information hierarchy, and AI explainability while continuously validating technical feasibility with engineering.

Once the experience matured, we built a high-fidelity click-through prototype and conducted formative usability studies with clinicians from multiple hospitals.

Observing real workflows helped us simplify interactions, improve information hierarchy, and increase confidence in AI-assisted recommendations.

Criterion-by-criterion evaluation UI Placeholder — swap in a high-fidelity prototype screenshot of the eligibility evaluation screen.

Outcome

The solution received highly positive feedback from clinicians.

Participants consistently highlighted the potential to significantly reduce the time spent evaluating patient eligibility while maintaining confidence in clinical decision-making.

The project was later showcased at the HLTH Conference in Las Vegas as part of Care Intellect's vision for AI-assisted oncology workflows.

What I Learned

This project reinforced an important principle for designing AI in healthcare:

Technology earns trust when it explains its reasoning—not when it hides complexity.

Rather than designing AI to replace expert judgment, we designed an experience that amplified clinical expertise while significantly reducing manual effort.