Inspiration
We addressed the challenge that many cancer patients and their families face in understanding complex and jargon-heavy clinical trial information, making it hard for patients to know which trials are right for them and make informed decisions about their participation.
What it does
Our search engine sources data from ClinicalTrials.gov and we filter their data based on our user's needs such as Gender, Age, Location and then provide all the options they have. Included is an AI powered summarizer that rephrases complex descriptions into easy-to-understand sentences for the user,
How we built it
We gathered recruiting U.S. clinical trial data from ClinicalTrials.gov, cleaned it, and standardized fields like age and location. We implemented a search function that filters trials by age, sex, state, and condition keywords. To make results easier to read, we used TF‑IDF scores with position weighting to extract key trial details into layman‑friendly summaries. The front end was built in Gradio with a search form, paginated results, and expandable “Read More” views.
Challenges we ran into
The raw dataset had inconsistent values for fields like age and location, requiring careful cleaning. Summarizing without losing important details was tricky, especially in medical contexts. We also had to handle pagination and UI state management in Gradio, which required creative workarounds. Performance optimizations were needed to keep the app responsive with large datasets.
Accomplishments that we're proud of
We made a fully functional search tool that helps patients quickly find relevant trials. The summaries make highly technical information easier to understand without removing key facts. Our interface lets users view detailed trial info without being overwhelmed. Most importantly, we did all this within tight hackathon time constraints.
What we learned
We learned how to adapt complex health data into something approachable for patients and families. Building a custom extractive summarizer taught us the strengths and limitations of TF‑IDF methods. We discovered how small UI tweaks like inline expanders can greatly improve usability. We also gained practical experience with state management in Gradio.
Log in or sign up for Devpost to join the conversation.