MCICT: Graph convolutional network-based end-to-end model for multi-label classification of imbalanced clinical text

Open Webpage

Yao He, Qingyu Xiong, Cai Ke, Yaqiang Wang, Zhengyi Yang, Hualing Yi, Qilin Fan

Published: 2024, Last Modified: 22 Jan 2026Biomed. Signal Process. Control. 2024EveryoneRevisionsBibTeXCC BY-SA 4.0
Abstract: Highlights•Introducing CoEAI-GCN module to address class imbalance, incorporating additional information and leveraging label co-occurrence for improved classification.•Proposing a novel end-to-end model, MCICT, designed specifically for enhancing the performance of multi-label clinical text classification.•Extensive experiments on real clinical datasets show superior accuracy compared to state-of-the-art models.