CURRENTLY I'M…

Learning Rive, one micro-interaction at a time

Reading an Agatha Christie novel (and guessing the culprit wrong)

Practicing Spanish, one Duolingo streak at a time

© 2026 Poorva Kulkarni

Built with love

Email

LinkedIn

CURRENTLY I'M…

Learning Rive, one micro-interaction at a time

Reading an Agatha Christie novel

Practicing Spanish, one Duolingo streak at a time

© 2026 Poorva Kulkarni

Built with love

Email

LinkedIn

GO BACK

TL;DR

Problem

Journey

Research

Insights

Success Metrics

Design Process

Solution

Reflection

ROLE

Product Designer

Led end-to-end screen flows

TIMELINE

Aug - Dec 2025

TEAM

6-person team

Mayo Clinic Physiscians

TOOLS

Figma, FigJam

Claude, Lovable

ROLE

Product Designer

Led end-to-end screen flows

TIMELINE

Aug - Dec 2025

TEAM

6-person team

Mayo Clinic Physiscians

TOOLS

Figma, FigJam

Claude, Lovable

TL;DR

Only 1 in 20 cancer patients joins a clinical trial, partly because finding one means decoding clinical language across sites that disagree. Mayo Clinic sponsored our six-person team to design a patient-facing alternative. I designed the end-to-end screen flows that connect patient and caregiver journeys, and shaped how the AI explains every match.
15+
Interviews
3 → 1
Concepts Merged
6 Screens
reworked after lo-fi testing
PROBLEM
The week after diagnosis
Sarah, 45, was just diagnosed with Stage III breast cancer.
She has two kids and a treatment plan. She also has a question her oncologist didn’t have time to answer:
could a clinical trial help?
“I don’t know what half these eligibility terms mean.”
This showed the need for a calmer, clearer path to the right trial.
JOURNEY
Understanding the real journey
The trials exist. Patients just can’t reach them.
From diagnosis to the doctor’s office, patients and caregivers hit a wall at every step. The hardest part sits
in the middle: searching, understanding eligibility, and comparing options.
Sarah wasn’t searching for a trial. She was decoding one.
RESEARCH
Finding the patterns
How we studied trial discovery
We combined affinity mapping, competitive analysis and targeted interviews with 15+ physicians. Every source pointed to the same root cause: the bottleneck wasn't a lack of trials, it was a lack of clarity.

COMPETITIVE ANALYSIS

Three sites, three answers

None explained why a trial actually fits.

AFFINITY MAPPING

Confusion, not lack of interest

Patients stalled on jargon and unclear next steps.

TARGETED INTERVIEWS

Clarity was the gap

Physicians, speaking for patients,

saw the same drop-off.

Inside the clinic
Mayo Clinic physicians set four ground rules:
01 PRIVACY

Let patients choose
who sees what.
02 MATCHING

Base every match on
real medical data.
02 MATCHING

Base every match on
real medical data.
03 RECORDS

Records are more
reliable than memory.
04 WORKFLOW

Leave the final call
to the doctor.
INSIGHTS
Synthesizing insights
Patients weren’t short on options. They were short on answers.
Patients didn’t need more trials. They needed reasons they could trust, and someone beside them.
01 SEARCH

Search is scattered across sites that disagree with each other.
02 ELIGIBILITY

Trust in AI is conditional. Patients need the why.
03 TRUST

Search is scattered across sites that disagree with each other.
04 CAREGIVERS

Caregivers do unpaid case management with no tool built for it.
05 EMOTION

The hardest part isn’t medical. It’s emotional.
SUCCESS METRICS
Defining Success
What we set out to improve
Our insights pointed to three directions. We explored them through three concepts, then merged the strongest parts into one product.
DESIGN PROCESS
Design direction
Three directions that shaped every screen

We merged three concepts into one direction, then I connected every patient and caregiver screen.

From three concepts to one
We merged the strongest parts of each concept into one product.

Think-aloud reviews and semi-structured interviews with physicians and faculty, before the final design.

Low-Fidelity prototype
The merged concept, clickable

The full patient journey in one prototype, built to test with physicians and faculty.

Low-Fi testing and iteration
What changed after testing

We ran think-aloud reviews and semi-structured interviews with physicians and faculty, then reworked five areas before the final design.

A direction we dropped
A complete intake, at the wrong moment

Low-Fidelity onboarding screen flows asked for a full medical history upfront. Physicians found it too long and daunting for someone who was just diagnosed. We cut it to the essentials and moved the deeper questions to TrialGuide, later in the journey.

SOLUTION
How we designed the AI
The AI suggests. The doctor decides.

Tradeoff: We chose a likelihood breakdown over a simple yes or no, because patients needed to see why they might qualify, not just whether.

Introducing TrialBridge
From confusion to confidence.

Think-aloud reviews and semi-structured interviews with physicians and faculty, before the final design.

A direction we dropped

Our high-fidelity onboarding asked for a full medical history upfront. Physicians found it too long and daunting for someone just diagnosed, so we cut it to the essentials and moved deeper questions to the TrialGuide chatbot.

REFLECTION
Looking back
Words are part of the design.

For someone facing cancer, one confusing word can stop them. Owning the flows taught me that most

of the complexity sits between screens.

If I did it again with AI

  • Pressure-test plain-language eligibility with an LLM before designing.

  • Map every likelihood state early.

  • Build in a “Did this make sense?” check from day one.

NEXT STEPS:
  1. Test with real patients
  1. Clinician dashboards
  1. Accessibility audit