摘要
本研究旨在提高婴儿培养箱不良事件报告完整性分类的效率和准确性,采用了融合Bert模型、双向门控循环单元(BiGRU)和多头注意力机制(Attention)的模型。研究使用某省提供的数据集,对报告进行细致分析,成功区分完整报告与不完整报告。实验结果显示,所提出的Bert-BiGRU-ATT模型在完整性分类任务中的平均F1值达90.37%,显著优于传统模型,证明了其在特定领域文本处理的有效性。该模型的应用将提升报告的可用性和准确性,有助于预防医疗事故,减少患者伤害,并减轻医疗专业人员的工作负担。
This study aims to improve the classification efficiency and accuracy of adverse event reports in infant incubators,using a model that integrates Bert model,bidirectional gated recurrent unit(BiGRU),and multi head attention mechanism(Attention).We conducted a detailed analysis of the report using a dataset provided by a certain province and successfully distinguished between complete and incomplete reports.The experimental results showed that the proposed Bert BiGRU ATT model achieved an average F1 score of 90.37%in integrity classification tasks,significantly better than traditional models,demonstrating its effectiveness in text processing in specific domains.The application of this model will improve the usability and accuracy of reports,help prevent medical accidents,reduce patient injuries,and alleviate the workload of medical professionals.
作者
李天纯
朱婉婷
夏文科
李维奇
张培茗
LI Tianchun;ZHU Wanting;XIA Wenke;LI Weiqi;ZHANG Peiming(School of Health Science and Engineering,University of Shanghai for Science and Technology,Shanghai 200093)
出处
《软件》
2024年第11期93-98,共6页
Software
