Medical Records Classification Model Based on Text-Image Dual-Mode Fusion
Yang Chen, Xinyue Zhang, Tao Li · 2021
With the advancement of science and technology and the improvement of policies, the construction and full realization of a paperless medical record filing management system in hospitals can have a positive impact on manpower, material resources, and financial resources. In order to better improve the accuracy and speed of medical record classification, A hybrid model based on text-image dual-mode fusion is proposed, which combines residual network (ResNet) and bidirectional gated recurrent unit (BiGRU). The fusion model realizes the fusion of text features and image features by splicing the output of ResNet and BiGRU. The experiment shows that the fusion model effectively improves the classification performance of a single model and has good application prospects.