Inspiration
Understanding blood test reports can be overwhelming for non-medical users. Reports are often delivered as dense PDFs filled with medical jargon, tables, and reference ranges that are difficult to interpret. We wanted to make these reports more accessible by transforming static medical documents into clear, readable summaries using AI.
What it does
Blood Report Summary allows users to upload a blood test PDF and receive an AI-generated, easy-to-understand summary. The system extracts structured text and layout from the document, highlights key findings, and presents the results in a clean web interface. The final output can also be deployed as a static web page for easy sharing and viewing.
How we built it
We built the system using a multi-stage pipeline:
PaddleOCR-VL is used to extract text, layout, and images from uploaded PDF blood reports.
The extracted content is converted into Markdown.
An ERNIE-based language model processes the Markdown and generates a structured, human-readable summary.
UI LINK:https://onevisionary.github.io/blood_report_analyser.github.io/
The summarized content is rendered into a clean web page.
The final output can be deployed on GitHub Pages as a static site.
Challenges we ran into
Handling complex PDF layouts with tables and medical values.
Managing large PDF files and API timeouts during OCR processing.
Ensuring the AI output remains informative without providing medical diagnoses.
Separating dynamic AI processing from static web deployment for GitHub Pages.
Accomplishments that we're proud of
What we learned
Document layout understanding is just as important as text extraction.
Clear prompt design is critical for generating responsible medical summaries.
Static site deployment requires a different architecture than dynamic web apps.
AI can significantly improve accessibility when used thoughtfully and responsibly.
What's next for Blood Report Summary
Highlighting abnormal values automatically using reference ranges.
Supporting multiple report types and languages.
Adding data visualizations for trends and key metrics.
Improving design and accessibility for wider adoption.
Exploring integrations with healthcare platforms while maintaining privacy and safety.
Log in or sign up for Devpost to join the conversation.