Improve the efficiency and accuracy of ophthalmologists' clinical decision-making based on AI technology

Abstract

Abstract Objective This article proposes a named entity recognition model for electronic medical records in ophthalmology that integrates professional vocabulary information. The aim is to achieve structured processing of important clinical decision-making data and to develop a clinical aided diagnosis platform based on this. The effectiveness of this platform in improving the efficiency and accuracy of ophthalmologists in clinical diagnosis decision-making was validated. Methods Based on the best entity recognition model, we constructed the aided diagnosis platform. By conducting a controlled experiment that compared the use of the platform by doctors with different levels of experience, we analyzed the effectiveness of the aided diagnosis platform in improving diagnosis decision-making efficiency and accuracy. Results The SoftLexicon-Glove-Word2vec model had the highest F1 score at 93.02%. Both junior and senior doctors showed significant improvement in diagnosis efficiency and accuracy (P < 0.05) when using the platform. Regardless of whether the aided diagnosis platform was used or not, there were significant differences in diagnosis decision-making efficiency and accuracy between junior and senior doctors (P < 0.05). Conclusion The use of artificial intelligence technology to construct the aided diagnosis platform for fundus diseases can effectively improve the clinical decision-making ability of junior doctors, and improve the diagnosis efficiency and accuracy.