MedLens

• 

  Banner for MedLens

Inspiration

Understanding medical reports can be difficult for non-experts. The complex jargon and lab results often leave people feeling anxious and confused. We were inspired to create a tool that helps people make sense of their health data in a way that is simple, private, and empowering. Our goal was to bridge the gap between raw medical data and actionable understanding, without replacing doctors—just making their patients more informed.

What it does

MedLens is a browser-based tool that transforms complex medical documents into plain-English summaries—now with text-to-speech and Spanish translation. Users can upload a medical report (PDF or image), and MedLens will:

• Extract and interpret the content using OCR and natural language understanding
• Translate the summary into Spanish for multilingual access
• Read the summary aloud for users who prefer or need auditory assistance
• Generate a list of personalized follow-up questions and export them as a downloadable file
• Include a symptom checker where users can input current symptoms and receive possible explanations or relevant concerns to discuss with their doctor
• Provide a downloadable summary PDF that users can bring to appointments or share digitally All of this runs entirely in the browser—no data is stored or sent to any server, making it private, secure, and HIPAA-conscious.

How we built it

We built the frontend using HTML, CSS, and JavaScript. For OCR, we used Tesseract.js to extract text from images. PDF.js was used to extract content from PDF files. We created modular JavaScript scripts to handle OCR processing, PDF parsing, text summarization, risk flagging, and PDF generation. We designed the application to run entirely in the browser to protect user privacy. The summarization and question generation logic is designed to optionally use large language models like GPT-4 through stateless API calls. Additionally, we used Hugging Face model for the translation of the text. No data is stored or sent to a backend server.

Challenges we ran into

One of the biggest challenges was ensuring HIPAA safety and protecting user privacy. We avoided storing or transmitting any data by keeping everything client-side. OCR accuracy was another issue, especially with low-quality scans or handwriting. Summarizing medical content in a user-friendly but safe way was also tricky—we had to carefully craft prompts and test many iterations. Lastly, handling inconsistencies in lab report formats made parsing difficult and required flexible design.

Accomplishments that we're proud of

We are proud that we built a fully functional prototype that can read, analyze, and summarize medical reports while keeping everything local to the user’s browser. We managed to integrate OCR, PDF parsing, summarization, risk analysis, and export functionality into a single, clean interface. We also created a HIPAA-aware system that avoids collecting or storing any health information, making it safe and trustworthy for public use.

What we learned

We learned how to integrate browser-based OCR and PDF tools, how to build modular JavaScript components, and how to prompt LLMs for medical summarization responsibly. We also gained experience with legal and ethical constraints around health data, user interface design for non-technical audiences, and building privacy-first tools. It was also a great exercise in making AI accessible and meaningful in real-world applications.

What's next for MedLens

We plan to develop a downloadable desktop version of MedLens that runs fully offline, ensuring complete privacy even without internet access. We're also working on expanding language support beyond Spanish to make the tool accessible to more users worldwide. The symptom checker will be enhanced with more personalized and context-aware guidance to provide better insights. To further improve usability, we aim to add voice-based interactivity, allowing users to speak their symptoms or questions directly into the tool. Accessibility is a top priority, so we’re also integrating support for screen readers and alternative input methods. Additionally, we’re exploring partnerships with telehealth platforms and HIPAA-compliant providers to offer optional cloud-based features. Ultimately, MedLens is more than just a translator—it’s a personal health companion that empowers users to better understand their medical data, ask informed questions, and feel confident in managing their health.

Built With

• css
• html
• huggingface
• javascript
• python
• tensorflow