Clinical Decision Support Prototype in EHR
How might we integrate predictive health data into a clinician’s existing EHR workflow without adding cognitive load or disrupting patient care?
Translational Innovation Program, MICHR
Healthcare
This project is under a non-disclosure policy.
Interface Design
1 year
UX Research
Usability Testing
Michigan Medicine
Context
To address delays in clinical intervention, a predictive risk assessment model was developed to flag at-risk patients in advance. Initially tested with a response team, the tool was envisioned to scale across different care settings to strengthen early intervention. To be effective, however, it required more than algorithmic accuracy: it needed a thoughtful blend of human-centered design and systems thinking to integrate seamlessly into the electronic health record (EHR). The design goal was to reduce clinician cognitive load and support timely decision-making.
A low-fidelity sketch of the interface
Adoption readiness
Participants reported higher trust and usability when
risk and protective factors were simplified in the UI.
My Role
User Experience Designer | Interaction Designer | User Testing specialist
Methods
Affinity Mapping | Semi-Structured Empathy Interviews | Thematic Analysis Sketching | Wireframing | Stakeholder workshops
Tools
Zoom | Miro | Google Workspace | Figma
EHR integration
Stakeholder Collaboration
Workshop Facilitation
Usability Testing Outcomes
Navigation efficiency
Clinicians completed critical patient data tasks significantly faster with the prototype compared to baseline workflows.
Validated design directions
Testing confirmed preferences around data display,
visual cues, and iconography, shaping refinements for future iterations.
Hello, World!
Interface insights
Usability sessions surfaced areas of confusion
e.g., data value displays) and highlighted opportunities for iteration prior to broader rollout.
Model functionality and scalability
It was highlighted that the model needed to be scalable to accommodate the distinct needs of various clinical units, such as adult and pediatric departments.
Role-based differences
Nurses, clinicians, doctors and hospitalists, each had unique preferences for the type of information displayed and how they process that information.
Difference in workflow on the floor
It became clear that nurses operated differently on the floor. They preferred to print the risk scores and assess patients by floor, starting from the topmost to the bottommost floors.
Key points
Reflections
Adapting to systemic constraints
During the presentation of the wireframes to the team responsible for integrating the model into the existing electronic health system, I encountered significant system constraints. This challenge forced me to pivot and adjust my design approach to align with the technical limitations, ensuring that the integration process would be seamless and efficient.
Translating qualitative insights to quantitative results
I primarily used qualitative methods and design thinking approaches, however, I needed to translate the findings into a format that could be understood by data scientists, researchers, and clinical stakeholders, ensuring that the results were actionable and accessible to all involved.
Embracing Project Shifts
After presenting the report to both my immediate team and the EHR system team, it was revealed that the project had been reprioritized and rescheduled for 2025. While not every project reaches its conclusion as initially planned, this shift offered an opportunity to reflect on the progress made and anticipate future adjustments.