In 2021, the Translator User Interface Team was formed to develop an interface for translational scientists and researchers to easily generate a manageable list of potential treatments for rare diseases and expedite the prescription of those life-changing therapeutics. Across hundreds of biomedical databases and journals there are often dozens of synonyms for a single term, each with a unique set of supporting evidence for thousands of known or potential treatments. However, many scientists are unable to use these other tools due to lack of experience with JSON and other computer programming languages, forcing reliance on their more tech-savvy colleagues to compose their research queries.

Once a query is run, researchers dig through thousands of results, massive graphs of how the results affect the selected disease, and large quantities of evidence before running the query several more times with multiple synonyms for each term. I demonstrated a series of wireframes to our shareholders and dozens of usability testing participants to discover what layout and workflow made natural language searching, filterable results, complex evidence, and graphs easy for biomedical researchers to understand, investigate, and trust.

After we defined the minimum viable product using what we had learned from my user research, I developed an atomic design system inspired by the existing NCATS branding with all of the components we would need to be able to build the live site, including graph nodes, popover styles, and filtering options. I then created a high-fidelity prototype with these components and conducted a usability test with translational scientists from a variety of biomedical research institutions.

In this first test, I discovered that users needed to see the publication source for each piece of evidence supporting a result rather than just the source ID or link. To increase users' trust in their results, I added filters for FDA approval status and type of source (clinical trial, publication, etc.). With these changes, I designed a second high-fidelity prototype that was used as a reference to our developers as they implemented a live version of the site for further usability testing.