本文为美国斯坦福大学（作者：Lucila Ohno-Machado）的博士论文，共256页。

近年来，尽管神经网络已应用于医疗问题，但由于各种原因，其适用性受到限制。其中一个障碍是稀有类别的识别问题。在本文中，演示并证明了一种解决这个问题的新方法的实用性。特别是开发了一种方法，对于稀有类别的识别具有高灵敏度和特异性，并将证明该方法是实用和强大的。该方法涉及序列神经网络的构造。如果要实现神经网络技术的实际应用，就必须学会稀有类别的处理。

生存分析是该类问题出现的一个领域。在这项工作中，检验了以下假设：（1）神经网络序列系统产生的结果（在校准和分辨率方面）比非分层神经网络更准确；（2）在某些情况下，序列神经网络产生的生存时间估计比Cox比例风险和逻辑回归模型更准确。使用了两组数据来检验这些假设：（1）一组HIV+患者数据集（ATHOS数据集）；（2）一组前瞻性随访的患者数据集（Framingham数据集）。利用ATHOS数据集，证明了神经网络模型比Cox比例风险模型能够更准确地预测艾滋病死亡。此外，还证明了序列神经网络模型比标准神经网络模型更精确。使用Framingham数据集，证明了逻辑回归和神经网络的预测没有显著不同，但是这些模型中的任何一个都比标准模型更精确。以序列方式使用预测模型进行生存分析是有利的，因为它能更好地利用现有信息。正如我在这项研究中所证明的那样，序列方式通常在不牺牲校准的情况下提高分辨率，也有助于为个人而不是为患者群体描绘相应疾病的进展模式。

Although neural networks have been applied to medical problems in recent years, their applicability has been limited for a variety of reasons. One of those barriers has been the problem of recognizing rare categories. In this dissertation, I demonstrate, and prove the utility of, a new method for tackling this problem. In particular, I have developed a method that allows the recognition of rare categories with high sensitivity and specificity, and will show that it is practical and robust. This method involves the construction of sequential neural networks. Rare categories occur and must be learned if practical application of neural-network technology is to be achieved. Survival analysis is one area in which this problem appears. In this work, I test the hypotheses that (1) sequential systems of neural networks produce results that are more accurate (in terms of calibration and resolution) than nonhierarchical neural networks; and (2) in certain circumstances, sequential neural networks produce more accurate estimates of survival time than Cox proportional hazards and logistic regression models. I use two sets of data to test the hypotheses: (1) a data set of HIV+ patients (AIDS Time-Oriented Health Outcome Study-ATHOS data set); and (2) a data set of patients followed prospectively for the development of cardiac conditions (Framingham data set). Using the ATHOS data set, I show that a neural network model can predict death due to AIDS more accurately than a Cox proportional hazards model. Furthermore, I show that a sequential neural network model is more accurate than a standard neural network model. Using the Framingham data set, I show that the predictions of logistic regression and neural networks are not significantly different, but that any of these models used sequentially is more accurate than its standard counterpart. The sequential use of predictive models for survival analysis is advantageous because it makes better use of the available information. It often increases resolution with no sacrifice of calibration, as I demonstrate in this study. It also helps to delineate patterns of disease progression for individuals, rather than for groups of patients.

1 引言

2 神经网络在医学上的应用

3 人工数据集中的稀有类别识别

4 用于诊断的分层神经网络

5 序列神经网络预测

6 评估方法

7 基于Framingham数据集的生存模型

8 HIV感染存活模型

9 结果讨论与结论

10 总结与未来工作展望 

下载英文原文地址：

http://page3.dfpan.com/fs/2lcja2215291a6145b8/ 

更多精彩文章请关注微信号：