首页> 外文会议>International Conference on Biomedical Engineering >Explainable and Actionable Machine Learning Models for Electronic Health Record Data
【24h】

Explainable and Actionable Machine Learning Models for Electronic Health Record Data

机译:用于电子健康记录数据的可说明和可操作的机器学习模型

获取原文

摘要

Advances in machine learning show immense promise for healthcare applications.Machine learning methods can accurately identify the relationships between input phenotypic variables and output clinical conditions,proving to be a useful tool for the prediction of clinical incidents [1].In particular,a suite of deep learning frameworks have been successful in predicting clinical incidents from electronic health record(EHR)data,as deep learning can represent complex,non-linear decision functions [2].Using de-identified EHR data from multiple healthcare databases,Google Health built a set of deep learning models that outperform traditional healthcare prediction tools by around 5% [3].To parse clinical notes,state-of-the-art natural-language processing(NLP)methods utilise long short-term memory networks(LSTMs)and attention mechanisms to parse clinical notes [4].Overall,machine learning shows strong promise for applications in the healthcare domain.
机译:机器学习的进步显示医疗保健应用的巨大承诺..可以准确地识别输入表型变量和输出临床条件之间的关系,证明是预测临床事件的有用工具[1]。特别是套件 深度学习框架已经成功地预测电子健康记录(EHR)数据的临床事件,因为深度学习可以代表复杂的非线性决策功能[2]。从多个医疗保健数据库,Google Health建立了来自多个医疗保健数据库的De-entedified EHR数据 一套深入学习模型,优于传统的医疗保健预测工具左右5%[3]。解析临床注意,最先进的自然语言处理(NLP)方法利用长短期内存网络(LSTMS)和 注意机制解析临床注意事项[4] .Overall,机器学习对医疗领域的应用表现出强烈的承诺。

著录项

相似文献

  • 外文文献
  • 中文文献
  • 专利
获取原文

客服邮箱:kefu@zhangqiaokeyan.com

京公网安备:11010802029741号 ICP备案号:京ICP备15016152号-6 六维联合信息科技 (北京) 有限公司©版权所有
  • 客服微信

  • 服务号