MediAssist

Extracted Data
Results and AI Summary

Inspiration

Doctors and patients often deal with lengthy, unstructured medical reports. We wanted to simplify this by building an AI tool that extracts key vitals and provides an instant, clear summary for faster understanding and diagnosis.

What it does

MediAssist reads uploaded medical reports (PDFs), extracts essential vitals like blood pressure, glucose, heart rate, and temperature, detects abnormalities, and generates an AI-driven summary highlighting the patient’s condition and key findings.

How we built it

We used Streamlit for the interface, PyMuPDF for text extraction from PDFs, and Hugging Face Transformers (BART model) for generating medical summaries. Regex-based logic was implemented to detect vitals and classify severity levels.

Challenges we ran into

Parsing inconsistent report formats from PDFs. Ensuring accurate extraction of vitals across varied templates. Managing model performance on limited compute (CPU deployment).

Accomplishments that we're proud of

Built a working AI app that converts medical reports into structured, actionable insights. Automated both vital extraction and clinical summary generation in real-time.

What we learned

Handling unstructured medical data efficiently. Integrating NLP summarization models into lightweight web apps. Designing doctor-friendly AI tools using minimal code and open-source models.

What's next for MediAssist

Integrate OCR for scanned reports. Add visualization dashboards for trends in patient vitals. Fine-tune the summarizer using domain-specific medical data.