Project Overview
Clinical trials are essential for advancing cancer treatment, yet 85% fail to recruit enough patients on time. This bottleneck doesn't just delay life-saving therapies—it costs trial sponsors between $600k–$8M per day in operational burn.
Our AI-powered platform, Clinspire, was designed to solve this dual challenge. By replacing dense medical forms with an accessible, conversational interface, we created a system that guides patients with empathy while optimizing enrollment velocity for the business.
The Challenges
Over 85% of trials fail to recruit on time, costing sponsors $600k–$8M per day in operational burn.
Complex eligibility criteria create "false positives," costing clinical staff approx. $1,200 per candidate in manual screening time.
Low digital literacy effectively locks out 85% of the potential patient pool, severely limiting the Total Addressable Market (TAM).
Dense medical jargon (e.g., "Metastatic") causes high abandonment rates, killing conversion among eligible patients who feel overwhelmed.
Distrust in "black box" AI creates privacy fears that reduce user adoption and long-term platform retention.
Navigating Our Biggest Constraint: Our biggest constraint was that Mayo Clinic’s privacy policies prevented direct access to patients. To adapt, we collaborated closely with their clinicians and engineers instead. This pivot turned a limitation into an advantage, allowing us to uncover patient pain points and critical workflow bottlenecks directly from the provider’s perspective.
User Needs
Business Goals
Research Insights
To diagnose the 85% failure rate, we triangulated data from Affinity Mapping (to isolate cognitive friction) and Competitive Analysis (to identify market gaps). Targeted Interviews then confirmed the root cause: the bottleneck wasn't a lack of patients, but a lack of process clarity.
Confusing Medical Jargon Drove Users Away
Qualitative analysis revealed that complex terminology on individual trial pages consistently confused users. Patients weren't dropping out due to lack of interest, but due to fear of answering incorrectly, leading to high abandonment ("false negatives").
We worked with clinical experts to build an AI-assisted "Plain Language" engine. By translating medical criteria into 6th-grade reading level explanations, we directly expanded the Total Addressable Market (TAM) for recruitment.
Lack of Personalized Matching and Poor Navigation
Benchmarking top platforms like ClinicalTrials.gov, TrialJectory, and Antidote revealed they function as static databases. They lack real-time personalization, forcing users to "guess" their eligibility. This high cognitive load slows down the journey.
We shifted to a Conversational Guided Intake. Asking one question at a time reduces perceived effort, which is projected to boost completion rates by 30–60%, directly accelerating enrollment velocity.
Overwhelming Eligibility Criteria
Stakeholder interviews confirmed that "false positives" (ineligible patients applying) were a massive drain on resources. Clinicians were spending expensive hours manually screening candidates who were overwhelmed by the requirements.
We prioritized redesigning the eligibility section with Automated "Met/Unmet" Logic. By visually simplifying the criteria and filtering leads early, the system saves an estimated $120k per trial in administrative waste.
The Aha Moment
As we spoke with clinicians, a clear pattern emerged: the issues they faced during trial matching consistently traced back to the patient side-unclear histories, inconsistent terminology, and incomplete comorbidity details.
That was the moment it clicked for us.
We didn't have two problems to solve.
We had one root problem: the patient onboarding experience.
Improving the patient flow would naturally improve data quality, reduce clinician workload, and strengthen the matching pipeline. Once we recognized this, we pivoted our focus entirely to the patient journey.
Ideation
We visualized the patient journey to uncover exactly where the experience was breaking down. By analyzing the different paths a user could take, we isolated the specific friction points causing abandonment-specifically the moments where users felt alienated by jargon or overwhelmed by eligibility logic. This map turned vague drop-off data into precise design targets.
Initial Wireframes
Given the barrier of complex medical jargon and unclear requirements, I sketched out low-fidelity concepts to help patients instantly simplify terms with AI, compare manual versus automated onboarding options, and easily assess their eligibility to prevent false starts.
Final Designs
Building on the validated wireframes, I refined the high-fidelity UI to establish patient trust. I prioritized a clean visual hierarchy that helps users instantly toggle AI simplification, verify their eligibility status, and complete the enrollment flow without cognitive fatigue.
We organized eligibility criteria into clear sections Met, Unmet, and Need Review, so users could easily understand their status and progress, effectively reducing confusion.
By simplifying medical jargon with AI and presenting clear eligibility criteria, we make information accessible to users with varying health literacy, helping them quickly understand their status.
Final Prototype
Outcomes
Clinspire’s design is expected to drive the following positive outcomes, making clinical trial participation more human-centered and equitable.
Projected outcomes are modeled on industry benchmarks from the Tufts Center for the Study of Drug Development (CSDD) and HHS Health Literacy data. These sources confirm that patient-centric automation can reduce recruitment timelines by 42% and save ~$1,200 per screen failure.
My Learnings
Empathy and flexibility are essential when designing for vulnerable populations.
Continuous engagement with stakeholders, especially clinicians and caregivers, leads to more inclusive and effective solutions.
Visual clarity and transparent AI explanations greatly improve user trust and comprehension.
Iterative prototyping and real-world feedback are key to refining digital health tools.
Future Enhancements
Due to strict medical privacy policies, I could not interview patients directly during this phase. My top priority for the next iteration is to bring real patients into the research loop, ensuring we validate our design decisions against their firsthand experiences and emotional needs.
