The problem
Hospital for Special Surgery is the #1 ranked orthopedic hospital in the US. But its clinical and research teams were working with fragmented data tools — spreadsheets, disconnected dashboards, and manual processes that slowed down decisions.
I joined as Design Lead for the Data Team in 2023 to help build a unified platform where clinicians, researchers, and analysts could query, explore, and act on complex patient data — without needing a data science degree.
Design leadership from strategy to ship
As the sole design lead on the data team, I own the full design process — from discovery and strategy through delivery and iteration. That means working closely with a product manager, engineering leads, and clinical stakeholders to define what gets built and how.
Synthesis from user interviews and workflow observation sessions with clinical researchers
Starting with the work, not the data
Clinical researchers don't think in SQL queries — they think in patient cohorts, outcome comparisons, and publication timelines. The first thing I did was spend time embedded with the research team, watching how they actually worked before touching any design tools.
That grounding shaped everything: the mental models we designed around, the language in the interface, and the decisions we made about what to abstract versus expose.
Low-fidelity exploration of query builder patterns
Concept testing with a clinical researcher, iteration 3
Key design decisions
Three decisions shaped the platform most:
Plain language over data terminology. We replaced technical database terms with clinical vocabulary — "patients" instead of "records," "conditions" instead of "diagnosis codes" — which reduced onboarding friction and support load significantly.
Progressive disclosure for complexity. Power users needed access to advanced filters and raw data export. New users needed a guided, forgiving experience. A layered interface let both groups work at their own level.
Saved views as a first-class feature. Researchers run the same cohort queries repeatedly — monthly outcome pulls, pre-publication checks. Making saved views persistent, shareable, and version-tracked turned a power-user feature into a team-wide workflow change.
The query builder in production — supporting cohort selection across 40+ clinical variables
Still shipping
This is ongoing work — the platform launched internally in early 2024 and continues to expand. Current focus areas include outcome dashboard templates for research publications and a self-serve onboarding flow for new research teams.
More detail is available under NDA.