ROLE · UX Designer | CATEGORY · Healthcare | TIMELINE · Jul 2025 – Aug 2025 | ORGANIZATION · Global TIES at UC San Diego
My role: I interviewed four EMS responders, mapped user flows, and created the landing page, hospital, and stroke-assessment screens. I also prototyped the last-known-well time input and ran guided usability tasks.
The Problem: High Pressure, Blind Decisions
This project was designed alongside Global TIES at UC San Diego, a team developing a portable, dry-node EEG cap to help EMTs differentiate ischemic from hemorrhagic stroke in the field. The device is still in development. As the UX team, our role was to design the interface responders would use to interpret and act on its output during transport, since existing diagnostic tools are too bulky for a moving ambulance, and a wrong call in the field often means a costly secondary transfer.
Three breakdowns defined the failure:
Uninterpreted Signals. Responders lacked tools to translate neurological data into an actionable signal under pressure.
Distance-Only Routing. Ambulances were routed to the nearest hospital, not the one equipped for the specific stroke type.
Blind Handoffs. Receiving teams got incomplete pre-arrival data, forcing them to react at the ER door instead of preparing in advance.
Field Validation: The Power of Intentional Friction
I interviewed four EMS responders and ran guided tasks using stroke and non-stroke images to test symptom capture, incident reporting, and hospital sharing. The moderated simulation used a scripted stroke scenario with four responders. The clearest failure was in the BE-FAST assessment: "Time," meaning last known well time, a critical metric for treatment eligibility, was represented by an unlabeled checkbox. When responders saw a patient wearing a watch, all four checked it, capturing nothing clinically useful.
I prototyped an explicit last-known-well time field to replace the checkbox. In follow-up tasks, 4 of 4 responders recorded it correctly, compared with 0 of 4 at baseline.
The Solution: Surfacing Actionable Clarity Under Pressure
Our team designed the interface around the decisions responders needed to make: identify the stroke type, select the right hospital, and prepare the receiving team. I mapped user flows and created the landing page, hospital, and stroke-assessment screens.
Interpreted Stroke Signal.
The interface concept presents a proposed stroke classification, ischemic or hemorrhagic, alongside a confidence score instead of raw EEG waveforms. This describes the intended interface, not validated diagnostic performance.
Designing for Uncertainty.
The BE-FAST checklist guides responders through each indicator without relying on memory, including the explicit time field rather than an ambiguous checkbox.
Capability-Based Routing.
The proposed routing flow would rank nearby hospitals by treatment capability and staff availability rather than distance alone. Live hospital-data integration and routing performance remain to be validated.
Pre-Arrival Handoff Package.
The proposed handoff packages patient history, symptoms, witness notes, and EEG classification for sharing with the receiving hospital before arrival. Live transmission and its effect on hospital preparation remain to be validated.
Outcome: Eliminating Hesitation at Handoff
Small scripted retest. All four responders correctly entered the last known well time on retest, compared with none in the initial task. This result concerns one input in a simulated task, not overall device accuracy.
Reduced hesitation. Once the checkbox was replaced, responders moved through BE-FAST without pausing over what "Time" meant.
Limitations and Reflection
This was a small, scripted evaluation with four EMS responders. The results provide design direction rather than clinical evidence. The dry-EEG device and companion app are still in development and have not been validated in live ambulance or hospital settings.
The BE-FAST checkbox made that lesson concrete. It looked right and worked as designed, yet captured nothing clinically useful, and all four responders misread it the same way. Designing alongside a real hardware team made us more deliberate about what earned space on the screen. If we continued the project, we'd validate the other side of the handoff: whether the pre-arrival summary actually helps receiving teams prepare faster.
Other projects
© 2026 Jeffrey Liang