J Shanghai Jiaotong Univ Sci ›› 2024, Vol. 29 ›› Issue (1): 131-140.doi: 10.1007/s12204-022-2499-1

• • 上一篇    下一篇

交叉特征深度编码网络预测儿童脓毒症

陈潇1,2，张瑞1,2，汤心溢1,2，钱娟3   

  1. （1.中国科学院大学，北京 100084；2. 中国科学院上海技术物理研究所；中国科学院红外探测与成像技术重点实验室，上海 200083；3. 上海交通大学医学院附属上海儿童医学中心，上海 200127）
  • 接受日期:2021-04-30 出版日期:2024-01-28 发布日期:2024-01-24

Prediction of Pediatric Sepsis Using a Deep Encoding Network with Cross Features

CHEN Xiao1,2 (陈潇), ZHANG Rui1,2 (张瑞), TANG Xinyi1,2 (汤心溢), QIAN Juan3∗ (钱娟)   

  1. (1. University of Chinese Academy of Sciences, Beijing 100084, China; 2. Shanghai Institute of Technical Physics of the Chinese Academy of Sciences; Key Laboratory of Infrared Detection and Imaging Technology, Shanghai 200083, China; 3. Shanghai Children’s Medical Center, Shanghai Jiao Tong University School of Medicine, Shanghai 200127, China)
  • Accepted:2021-04-30 Online:2024-01-28 Published:2024-01-24

PDF (PC)

29

摘要： 脓毒症严重威胁儿科重症监护病房儿童的健康。通过及时诊断和治疗干预，可有效降低小儿脓毒症的死亡率。杆菌培养检测法耗时长，无法及时治疗。提出了一种新的框架：一个具有交叉特征的深度编码网络（CF-DEN），可以准确地进行早期脓毒症的检测。通过梯度提升决策树自动构造交叉特征，并提取到所设计的深度编码网络（DEN）中。DEN旨在从临床试验数据中学习足够有效的表征。DEN的每一层通过注意力机制对当前层计算中涉及的特征进行过滤，逐层叠加输出当前预测，得到最后一层嵌入特征。该框架利用基于树的方法和神经网络的方法，从小型临床数据集中提取有效表征，得到准确预测，从而促使患者得到及时治疗。在上海儿童医学中心收集的数据集上评估了框架的性能。与常见的机器学习方法相比，我们方法在测试集上的F1-score提高了16.06%。

关键词: 儿童脓毒症，梯度提升决策树，交叉特征，神经网络，交叉特征深度编码网络

Abstract: Sepsis poses a serious threat to health of children in pediatric intensive care unit. The mortality from pediatric sepsis can be effectively reduced through in-time diagnosis and therapeutic intervention. The bacilliculture detection method is too time-consuming to receive timely treatment. In this research, we propose a new framework: a deep encoding network with cross features (CF-DEN) that enables accurate early detection of sepsis. Cross features are automatically constructed via the gradient boosting decision tree and distilled into the deep encoding network (DEN) we designed. The DEN is aimed at learning sufficiently effective representation from clinical test data. Each layer of the DEN filtrates the features involved in computation at current layer via attention mechanism and outputs the current prediction which is additive layer by layer to obtain the embedding feature at last layer. The framework takes the advantage of tree-based method and neural network method to extract effective representation from small clinical dataset and obtain accurate prediction in order to prompt patient to get timely treatment. We evaluate the performance of the framework on the dataset collected from Shanghai Children’s Medical Center. Compared with common machine learning methods, our method achieves the increase on F1-score by 16.06% on the test set.

Key words: pediatric sepsis, gradient boosting decision tree, cross feature, neural network, deep encoding network with cross features (CF-DEN)

中图分类号: