devicewise: A Medical Device Safety Intelligence Platform

Framing the Problem

Patients can easily access information about their medications, including prices, side effects, and drug interactions. Yet when it comes to implantable medical devices, transparency is limited, leaving patients excluded from decisions about devices that may remain in their bodies for years and have lifelong consequences. Moreover, several platforms do exist which provide comprehensive information on medications, but these are largely tailored to the US healthcare system and exclude implantable medical devices.

Idea Explanation

What is your idea? How does it fix the problem?

devicewise is a web and mobile platform for the UK market that brings together medical device information into one accessible place. Rather than searching across multiple sources, users can view:

• Device/manufacturer specifications

• Comparisons between similar devices

• Clinical research with AI-generated summaries

• Reported safety incidents and recalls

• Clinical trial information

Our aim is to make reliable medical device information more transparent and easier for patients to understand, while also providing a useful reference for healthcare professionals.

Implementation

We produced the frontend using Loveable which used mock data.

Our proposed architecture would be the following:

• An ETL pipeline that collects and organises information from multiple sources.

• A central database that stores the cleaned device information.

• An AI layer that generates concise summaries of scientific literature.

• An API that delivers the processed data to the web and mobile application.

For this prototype, we came up with an aggregation of the following data sources: mock data, live API data and scraped datasets, reflecting the fact that many relevant healthcare sources do not currently provide publicly accessible APIs.

Challenges

What did you struggle with? How did you overcome it?

The biggest challenge was data availability. While for example the ISRCTN (for trial clinical trials) did have a public API, other medical device related websites did not. To overcome this, we would build a hybrid architecture for a mockup that combines mock data, live API connections and web-scraped data. This allowed us to demonstrate the platform's functionality while creating an architecture that can be expanded as more data sources become available.

Accomplishments

What did you learn? What did you accomplish?

Through this project we gained a deeper understanding of the lack of transparency surrounding implantable medical devices and the challenges of integrating data when there is no public API, which meant we had to think of other ways of collection data. In summary, this project allowed us to learn about a new healthcare domain, strengthen our technical skills, and practice with developing a comprehensive solution within a limited timeframe.

Next Steps

What are the next steps for your project? How can you improve it?

Next steps would focus on expanding the data pipeline by integrating additional APIs where possible and have an automated ETL process to continuously update the database with new information (e.g. clinical studies, safety alerts, regulatory information etc.). Moreover, we were not able to connect the proposed API yet, or any of the data but this would be a goal for the next steps (first steps of how we would do it in github/Supabase)

We also plan to improve the AI summarisation layer, and expand the range of supported medical devices. Lastly, we would want the the possibility to specify side effects by gender and ethnicity in the app/website.


Thank you,
Fernanda Bueso, Suna Choi, Elise Brienne