Researchers release MeddeX: a LLaMA-3 medical report analyzer with Gemini Vision
Researchers introduced MeddeX, a retrieval-augmented system combining a fine-tuned LLaMA-3 (8B) model, Retrieval-Augmented Generation (RAG), and Gemini Vision for analyzing medical reports. It processes text-based PDFs, scanned images, and radiological scans while filtering corrupted files via a garbage-text detection mechanism. The system also compares longitudinal reports and processes legacy…
Key points
- MeddeX merges LLaMA-3 (8B) with RAG and Gemini Vision for medical report analysis, handling PDFs and scans
- System includes garbage-text detection, longitudinal report comparison, and on-demand legacy document processing
- Outperforms LLaMA3-Med42-8B by +9.2% to +71.4% across ROUGE, BLEU, and BERTScore metrics
MeddeX outperforms the base LLaMA3-Med42-8B model across benchmarks: ROUGE-1 (+9.2%), ROUGE-2 (+10.0%), ROUGE-L (+11.8%), BLEU (+71.4%), and BERTScore (+0.8%). Built with FastAPI, Supabase, and Next.js, it aims to reduce clinician workload by automating evidence-backed summaries.
The headline, key points and digest above were generated by Digest AI's editorial model from the linked sources. Automated summaries can contain errors: the sources are the record. Spotted a mistake? Tell us. Published by Martin K., who runs Digest AI and handles corrections.
More in Research
All →- OpenAI offers U.S. university labs free ChatGPT Business seats with Rosalind access · 2 src
- Anthropic reports small AI models may self-replicate on IoT devices within months · 29 src
- Researchers propose LLM-referenced C-test for evaluating supra-intelligence · 8 src
- Researchers study permutation diffusion LLMs for recommendation reranking · 1 src
- Researchers build self-improving AI agent that invents its own operators for exact answers · 1 src
Comments
via GitHub Discussions