When minutes cost neurons, the diagnostic gap is a data gap.
Stroke is the 4th leading cause of death in the United States, with roughly 800,000 cases each year. About 87% are ischemic — clots that are highly treatable when caught early. But in the field, up to 22% of strokes are missed or misdiagnosed by EMS. Every minute a stroke goes untreated, the brain loses 1.9 million neurons.
The diagnostic gap isn't a knowledge gap — it's a data gap. EMTs make critical, time-pressured calls without access to the patient's history or any objective brain data. That uncertainty costs minutes, and minutes cost lives.
Three pieces of evidence, one decision view.
Banambulance is an EMS-facing app that pairs three pieces of evidence to support faster, more confident stroke calls in the field.
- Real-time EEG output from a portable cap, surfacing objective brain abnormalities.
- Voice-driven patient history capture, so the EMT can keep their hands on the patient.
- The existing BEFAST behavioral protocol, restructured into a clear, time-aware UI.
“Together, these turn a high-pressure judgment call into a decision backed by data — and route the patient to the right facility, not just the nearest one.”
Two EMTs, one stroke survivor's mother, three findings.
I interviewed two practicing EMTs and the mother of a stroke survivor. From the conversations, three findings reshaped our direction:
- BEFAST is the protocol EMTs already trust. Building over it — not replacing it — was non-negotiable.
- EMTs aren't always confident in their behavioral assessment. Stroke-mimicking conditions and "silent" strokes are common, and a wrong call costs minutes downstream.
- Patient medical history is the missing context. With it, an EMT can disambiguate whether a symptom is acute or pre-existing.
Marcus & Aurora
- Marcus, 32 — EMT, 9 years experience. Comfortable with portable EEG hardware but finds existing readouts cluttered and slow to parse under stress. Wants confidence, not noise.
- Aurora, 35 — Charge Nurse. Needs a heads-up before patient arrival so she can spin up the right team and the right facility. Lack of incoming data forces her to prepare for too many possibilities.
From parallel screens to a single decision view.
We mapped the EMT's flow against the BEFAST checkpoints, then layered in EEG and patient history as supporting evidence rather than parallel screens. Early wireframes were tested with one of our EMT interviewees, whose feedback drove the largest iteration: collapse hospital recommendation, brain-data summary, and BEFAST output into a single decision view, so the EMT can radio the hospital with one screen open.
Empathy wasn't a step — it was the filter.
Working alongside neuroscientists and designers turned stroke misdiagnosis from a statistic into a problem I could feel the weight of. That changed how I made decisions — empathy with the user wasn't a step, it was the filter.
I also learned how tightly content strategy and visual design have to be coupled for a high-stakes, low-attention interface. A clean visual is irrelevant if it misorders the information. A good information hierarchy is irrelevant if the visual doesn't make it scannable. Making both serve the same beat — that's the skill I'll carry forward.