论文部分内容阅读
针对临床随访资料的多元分析方法种类多且各具特色的现状 ,为明确各种生存分析方法的特点及适用范围 ,在查阅中外文献和应用研究基础之上 ,进行总结分析 ,并对发展中的 BP神经网络在生存分析中的性能进行了模拟。结果表明 ,各类方法都有独到之处 ,但也存在需解决的问题 ;方法的选择取决于资料的类型和研究目的。因此 ,针对具体问题 ,要明确研究目的 ,分析资料特点 ,选择合适的模型 ;传统方法难以解决的非线性问题有望在神经网络模型中得到较好的解决
The multi-analysis methods for clinical follow-up data are diverse and have their own characteristics. In order to clarify the characteristics and applicable scope of various survival analysis methods, they are based on the review of Chinese and foreign literature and applied research, and are summarized and analyzed. The performance of BP neural network in survival analysis was simulated. The results show that each method has its own uniqueness, but there are also problems that need to be solved; the choice of method depends on the type of data and the purpose of the study. Therefore, for specific problems, we must clarify the purpose of the study, analyze the characteristics of the data, and select the appropriate model; nonlinear problems that are difficult to solve with traditional methods are expected to be better resolved in the neural network model.