Research and product strategy: visualizing oncology patient data
Overview
Problem. With little time and disparate systems, medical oncologists struggle to make sense of patient data. Oncologists rely heavily on self-reporting, but patients frequently forget or omit important between-visit events. Medical oncologists and patients need data tracking methods that help them maximize in-office interactions.
Solution. My team explored medical oncology workflow pain points and patient challenges and created web and mobile applications that offer an aerial view of patient treatment journeys.
Timeline
7 months
Role
Team lead, UX researcher
Team
5-person collaboration with 1 visual designer, 1 researcher, 1 interaction designer, 1 engineer
Skills
Project management, qualitative research (interviewing, script writing, moderation), workflow mapping
Understanding the healthcare landscape
To better understand the problem space, we created a territory map (from the perspective of our stakeholders) and an oncology touchpoint map (from the perspective of our users) highlighting the individuals and systems involved in the cancer care process. These visuals allowed us to develop both macro- and micro-level views of care delivery interactions.
Diving into primary research
We interviewed 13 medical oncologists at medical centers across the United States to get a better sense of how cancer care teams share information and collaborate. We took a bottom-up approach to synthesizing our findings, which allowed us to identify salient patterns and brainstorm data-driven solutions.
Key findings
01 - Electronic medical record (EMR) data is nested and in plain text, making information retrieval tedious for medical oncologists.
02 - Medical oncologists struggle to obtain a longitudinal, holistic view of each patient, which undercuts their treatment decision-making confidence.
03 - Patient quality of life (QOL) measures and symptom tracking tools are underutilized in cancer care.
04 - Patient self-report errors and misconceptions about side effects make it difficult for care team members to provide timely treatment adjustments.
💡Insight: Pressure to make the most of limited face-to-face time and fragmented data are the biggest contributors to negative oncology patient experiences.
🚩 Impact → Focus on collecting data between visits and facilitating more meaningful in-office conversations between providers and patients.
Ideating through sketching
We sketched a system that would offer medical oncologists an overview of patient status with workflow apps for patient side effect information, treatment response correlations, and regional outcomes. Our design also incorporated a patient companion app that collects between-visit information on side effects and QOL. This real-time data stream would help care teams prioritize and proactively address patient complaints.
Converging on an early prototype
Sketching allowed us to quickly recognize that segregated workflow apps would not resolve the issue of data inaccessibility. With our mid-fidelity prototype, we worked on consolidating our design into a single view that would disclose the right data at the right time.
Medical oncology dashboard
Patient companion application
Testing our solution with oncologists and patients
We conducted moderated, remote usability tests of our initial dashboard and companion application prototypes. In total, we tested with 11 medical oncologists and 3 patients.
Research insights
01 - Medical oncologists distrust artificial agents. While intrigued by the concept of an artificial intelligence (AI) agent that could surface hidden connections between treatment response variables, the majority of medical oncologists reported that they wouldn’t rely on these correlations over their own judgement.
02 - Caseload management matters as much to medical oncologists as individual patient management. We had failed to carefully consider our solution’s entry point.
03 - Patients and medical oncologists sacrifice privacy for real-time support. Through our testing, we uncovered the fact that many patients seek advice about side effects by sending texts to their medical oncologist’s personal smartphones.
Design impact
01 - Concerned about adoption, we removed the AI concept from the dashboard. Instead, we vertically stacked the treatment timeline, cancer profile, and subjective patient response graphs and added dotted treatment change lines, which allowed medical oncologists to more readily find patterns on their own.
02 - To help medical oncologists get oriented for their often hectic days, we added a home screen that displayed overall caseload and treatment anomalies from last visit.
03 - We incorporated a photo feature in our companion app to allow patients to privately and securely send images viewable by care teams in their desktop portal.
Finalizing a high-fidelity prototype
Our final solution included three vantage points that allowed medical oncologists to drill down from caseload overview to patient treatment timeline to individual test results.
Home: caseload view depicting patient names and demographics, last visit date, current treatment, and last reported side effect
Treatment timeline: longitudinal data visualization of patient’s medications, procedures, test results, and treatment changes
Test result modal: detailed view of latest bloodwork findings
Reflecting on the project experience
Though it has been nearly 5 years since I worked on this project, I keep it in my portfolio to remind myself of how far I’ve come and what I continue to value as a researcher, designer, and leader.
Be inclusive and encourage design participation
Instead of wallowing in obscurity, make ideas tangible and bring them out into the open. Even better, include users directly in the design process. Once we started showing functional creations (paper can work just as well as a digital prototype!) to our intended users, and co-designing with them, our learning soared.
Talk about hard things
My team struggled for the first few months of this project. With varying degrees of experience in the healthcare space, different team members faced different levels of ambiguity. Once we started spending extra time together to work through problems and approaching obstacles with more patience and curiosity, our team dynamic shifted for the better.