Melanoma Clinical Trials
Capstone Project
OVERVIEW
The process of matching patients with melanoma to suitable clinical trials is inefficient, time-consuming, and inaccessible for oncologists. The main website for searching clinical trials, Clinicaltrials.gov, only serves as a reporting tool and does not prioritize the best interests of patients.
To address this issue, we collaborated with Dr. Bryan Carroll to undertake a redesign of Clinicaltrials.gov. The redesign includes an enhanced search functionality, easily understandable trial details, and the ability to apply directly to trials. These improvements aim to provide oncologists with better-curated trial recommendations, ultimately saving them valuable time.
MY ROLE
UX Researcher, UX Designer
DURATION
January 2023 - May 2023
CLIENT
Dr. Bryan Carroll, University Hospitals of Cleveland
TEAM
Saloni Gandhi, Namita Rao, Sruti Srinidhi, Wenqing Yin
Understanding the problem
To better understand the problem space of why matching melanoma patients to appropriate clinical trials is a lengthy and cumbersome process, we conducted domain research, competitive analysis, and nine structured interviews. Our research goals were:
Research goal #1
Obtain a current state understanding of how patients get connected to clinical trials
Research goal #2
Identify the gaps in the current tools for finding clinical trials
Domain Research
We did extensive research to better understand the domain and process involved in melanoma and clinical trials. I took charge of creating shared documents with important terminology and findings to help establish a shared understanding with oncologists during interviews.
Competitive Analysis
The second part of our approach was to explore current resources that help match patients with clinical trials in order to understand the current solutions available and what they lack.
Structured Interviews
The third part of our approach was to talking directly to 9 dermatologists and oncologists to understand their current process of matching patients and their pain points. We used a storytelling approach with 5 mock patients we to help guide the interviews.
Affinity diagramming the interview results and domain research helped uncover key insights
Clinical trials are inaccessible to oncologists
The only resources to access clinical trials are through word of mouth or on clinicaltrials.gov.
Heavy reliance on personal connections
Relying on their internal network does not always offer the best care for the patient’s needs.
Determining patient eligibility is time-consuming
It can take multiple months to get in contact with a trial coordinator and enrolling a patient into a clinical trial.
The Challenge:
How might we increase the visibility of different approaches to oncologists so they can determine the right clinical trials and medical centers for melanoma patients faster?
Brainstormed 40 ideas and narrowed down to 4 solutions
Utilizing the "crazy 8's" technique, we generated a comprehensive set of 40 diverse ideas encompassing both physical and digital solutions.
Through a collaborative brainstorming process, we identified 12 promising concepts, highlighting their concerns or recognizing areas where significant value could be generated. After evaluating each one, we refined the list to set of 4 final ideas. We decided to storyboard each one to help visualise each concept when presenting to our client.
Idea 1
Facilitating referrals for clinical trials
Idea 2
Portal for standard application
Idea 3
Optimising search and filter
Idea 4
Online tumor board
Speed-dated storyboards led to pivot in scope direction
To gather feedback on the feasibility of each of the four ideas, we conducted speed-dating sessions with 7 oncologists and our client, presenting them with each storyboard. Things like how much extra time our tool would add to oncologists were discussed. In these discussions, we decided that improving Clinicaltrials.gov would prove to be more valuable than creating a standalone platform from scratch.
In-depth analysis of clinicaltrials.gov to locate origin of issues
After redefining our scope, we decided to analyze Clinicaltrials.gov, a federal government website that people use to find clinical trials. The goal was to better understand the issues and reasons behind oncologists' feedback during interviews, where they noted that it was inefficient, time-consuming, and inaccessible
The limited search criteria makes it hard to find relevant clinical trials
Accessing and engaging with the clinical trial is difficult
Accessing and engaging with the clinical trial is difficult
Ideated a new solution based feasibility, user-centeredness, and value proposition
Addressing the concerns the team found during analyzing Clinicaltrials.gov and the insights we discovered through the storyboard speed-dating sessions, we generated a new and innovative solution that demonstrates high value for oncologists, ensuring alignment with their needs and preferences.
Tested low-fidelity storyboard concept
The primary goal of user testing was to get feedback on our proposed storyboarded solution. We wanted to understand if our target users, oncologists, would truly find it helpful and if they had any thoughts or concerns. We were able to test our prototype on 6 different oncologists and receive their feedback. Based on our user testing, we arrived on three main insights.
1. The solution does help oncologists search for clinical trials and apply to them for their patients.
“100% I would use this platform”
“[This solution] would definitely target the issue we face”
2. Oncologists are busy and will not spend a long time to fill out the application.
“Doctors don’t have much time, won’t spend much time on something that is complicated”
“Would be even nicer if we could automatically extract necessary data from patient file”
3. Oncologists need to be able to trust the clinical trials recommender system.
“Show me the trials that did not match all criteria”
“Very important to also get similar trials and not just something local that they can just send them too”
Created a user flow
After validating the proposed solution would solve the problem, we went on to create a user flow diagram to help navigate and organize the creation of our mid-fidelity prototype.
Explored diverse design opportunities through parallel prototyping
Each team member was tasked with designing an interface for each of the workflows (Search, Recommended Trials, Trials Details, and Application). Through parallel prototyping, we were able to explore diverse visualizations of interactions and information. Presented below is the flow I personally designed, accompanied by insights into my design decisions:
Search for a clinical trial for a specific patient
-
Accessible on clinicaltrials.gov to avoid users switching back and forth between platforms
-
Patient data drawn from Electronic Medical Records to save time inputting information like past medical history
-
Ability to always go back to a regular search
Employed think-aloud protocols to test mid-fidelity prototypes
In order to gather feedback on the user workflow and interactions, we conducted testing of the mid-fidelity prototype with six different oncologists. Our objective was to employ think-aloud protocols to capture their reactions, questions, concerns, and overall impact. The testing process yielded the following findings:
-
Discrete searching options help narrow down and eliminate unfit clinical trials
-
Categorizing each trial's details improves scannability and saves time
-
Not all Electronic Medical Records are the same for each hospital which may make it hard to extract them from the search
Iterated prototype based on user feedback to create a high-fidelity clickable prototype
Design system to promote consistency across interfaces
Using the design system to create the interfaces
Search for a clinical trial for a specific patient
The new Advanced Search tab allows oncologists to search for clinical trials for specific patients. This provides them with search results that are more curated for their patient's conditions compared to their current search function.
The Advanced Search is organized into categories that grow in specificity, allowing oncologists to better navigate and easily answer the search questions.
Recommend clinical trials best suited for patient
Through the machine learning algorithm our team developed, oncologists can now see how much of a match a clinical trial is given the information they provided in the search. This allows oncologists to clearly see which clinical trials may be a match at first glance, saving time determining this manually.
The additional columns "Criteria Match" and "Criteria Non-Match" lets oncologists be able to see why or why not a clinical trial is a match. During testing, oncologists noted that this feature would save them time and effort in discovering this information themselves.
View clinical trial in detail
The current trial details page is very text-heavy, containing many long lists of text, and has been noted to be difficult to navigate by all oncologists interviewed. By organizing the text into categories, oncologies can easily scan the information they are searching for. A horizontal menu has been created to allow oncologists to easily jump to specific sections without having to scroll to find a section.
The Eligibility Criteria section has turned into an interactive checklist to assist oncologists in efficiently determining whether their patient is eligible for a clinical trial. The Eligibility criteria have been split into "Priority" and "Preferred" to further help oncologists digest information easier.
Oncologists often compare clinical trials to see which one would best suit their patient, which involves a lot of back and forth. A new Compare Panel has been added to help improve the efficiency of this process.
Ability to apply to clinical trials directly
Currently, you cannot apply directly to clinical trials on clinical trials.gov. To mitigate the month's oncologists spend getting in contact with the trial coordinator, oncologists can now apply directly on the website.
Multiple clinical trials can be applied at the same time with a single application, further saving time from oncologists doing it manually. The application is pre-generated using information from the oncologist's account.
Lastly, oncologists can submit their Match Percentage score to showcase to the trial coordinator that the patient will be a good fit for their trials, further saving weeks doing this manually over email and phone calls.
What is next?
Pitch solution to clinicaltrials.gov
The clinicaltrials.gov modernization team is dedicated to redesigning clinicaltrials.gov to be more usable. Our client will take over our designs and pitch the solution for them to implement into clinicaltrials.gov.
Scale across all cancers beyond melanoma
The solution has been made specifically for melanoma but is applicable and has the opportunity to be designed for all diseases beyond melanoma.
Fully build out the ML algorithm
The current ML algorithm is a proof of concept and only uses a few user inputs. Building it fully out will further filter down the clinical trials and provide oncologists with the best recommendations.
My takeaways
⏰
One of the biggest obstacles our team faced was reaching stakeholders. Their demanding schedules required us to not only effectively manage our own schedules to find suitable meeting times but also learn how to maximize the limited time we had with them. This challenge became particularly apparent when we had only 15 minutes for usability interviews. However, this experience taught us to carefully plan and prioritize our objectives for each meeting, ensuring that we could achieve productive outcomes for both parties involved.
The nature of the domain we worked in presented us with an incredible learning challenge. To address this, I took the initiative to create a comprehensive folder that captured all the technical vocabulary related to melanoma and clinical trials that we needed to know. This shared resource proved to be invaluable when interpreting oncologists' responses during interviews. By establishing a strong knowledge foundation and fostering shared understanding between the oncologists and our team, we were able to design a more impactful solution with greater ease.
✍🏻