Medy-Scan

Inspiration

Health records are scattered across paper documents, PDFs, and multiple clinic portals, making it difficult for patients to keep track of their medical history or understand what their results actually mean.

Nearly 90% of Canadians adults over 65 struggle to understand basic health information, and about 1 in 20 patients experience preventable harm in healthcare due to errors or missing information. In some cases, poor communication or incomplete records contribute to medication mistakes and missed treatments that could have been prevented with clearer information and better organization.

What it does

Medy-Scan is an AI-powered personal health record system to help people understand and manage their medical information. Users can scan or upload lab reports, prescriptions, and clinic notes using their phone, and the system analyzes the document to extract key medical information. The app identifies important values, and generates clear, patient-friendly explanations to improve health literacy.

All captured records are automatically organized into a chronological health timeline, allowing users to see their medical history in one place and track how different results or prescriptions change over time. The platform helps users stay more organized and maintain a clearer picture of their overall health by centralizing previously scattered documents into a single system.

Medy-Scan also includes an AI health assistant that can answer questions based on the user’s stored records. The assistant is able to explain lab results, highlight potential trends, and provide context around prescriptions or lab values. This allows users to interact with their health information conversationally instead of searching through complicated technical documents. The more you use Medy-Scan the better the system becomes, as it continuously builds a profile around your medical history..

How we built it

We built Medy-Scan as a cross-platform app using React Native with Expo, which allowed us to create one codebase that runs on iOS, Android, and the web. We wrote the app in TypeScript to keep the code organized and reduce errors as the project grew.

For the AI features, we used Vultr Cloud AI to run a 34-billion parameter model that analyzes medical documents. When a user uploads or scans a document, the model processes the image and extracts important information including test values and prescription names, and then generates a simple explanation so users can understand their results.

The backend and database are powered by Firebase, where we store user accounts and encrypted health records securely. Firestore allows us to organize timeline events and extracted data so that users can easily track their health history.

For authentication and security, we integrated Auth0, which helped us implement encrypted login and protect sensitive medical data. This ensures that each user’s health records remain private and only accessible to them.

Challenges we ran into

Building Medy-Scan had both design and technical challenges. Making the user interface intuitive was hard because medical records are complex, and we wanted a timeline and summaries that were easy to read. We took inspiration from several other apps before settling on our clean, simple design.

Picking a name also took a long time. We explored Greek mythology names, but none explained what the app actually does. Eventually, we chose Medy-Scan because it clearly communicates the purpose.

On the AI side, our 90-billion parameter Vultr model was very slow at first. We had to optimize it by adjusting its size and streamlining processing so users could get fast results without losing accuracy. Making all this work while keeping data private and reliable was a big challenge, but worth it to solve.

Accomplishments that we're proud of

We’re proud of several achievements with Medy-Scan. By using a 34-billion parameter AI model through Vultr, we were able to accurately analyze complex medical documents and power our insight engine while maintaining speed and privacy. After many iterations, we finalized a user interface that is simple, intuitive, and works seamlessly across mobile and web.

Even our simplest feature, scanning and verifying prescriptions, has the potential to prevent misdoses and save up to 24,000 Canadians each year. Overall, we’re proud creating a fully functional app that makes health data understandable and actionable, all in under 36 hours.

What we learned

Working on Medy-Scan was a huge learning experience for our team. One teammate got their first experience working with a database, gaining hands-on skills with structured data and Firestore. We also learned how to leverage cloud computing to scale our AI workflows and make processing large medical documents efficient. Finally, we gained practical experience with encryption and secure authentication through Auth0, reinforcing the importance of privacy and security in handling sensitive health data.

What's next for Medy-Scan

We plan to scale Medy-Scan by using bigger and more specialized AI models to improve accuracy and speed, and by expanding our databases to handle larger volumes of medical records. Future features will go deeper, including advanced trend analysis, prescription reminders, and multi-user support for families or caretakers. These improvements will make Medy-Scan even more powerful, proactive, and life-saving.

Built With

• auth0
• firebase
• react-native
• typescript
• vultr