Paper Title:
Medical Knowledge Infusion and Multi-Stage Correction Make LLMs Better Proofreaders for Radiology Reports
This repository contains the official implementation of our AAAI AISI 2026 paper.
Our work introduces a dual-stage framework that leverages medical knowledge infusion and multi-stage correction to enhance the proofreading performance of large language models (LLMs) on radiology reports.
Clone this repository and install dependencies:
git clone https://github.com/knowlab/MedKIC-Dual-Knowledge-Infusion-Framework-for-Radiology-Report-Proofreading.git
cd MedKIC-Dual-Knowledge-Infusion-Framework-for-Radiology-Report-Proofreading
pip install -r requirements.txt