双区间删失数据下基于 Stochastic EM 算法的比例优势模型的估计研究

Estimation of proportional odds model based on Stochastic EM algorithm under doubly interval censored data

  • 摘要: 潜伏期是流行病学、疾病进展研究等关心的重要指标之一, 对疾病防控及治疗具有重要作用. 潜伏期是从病毒感染到产生症状这两个事件发生时间的间隔时间, 并且这两个发生时间均有可能出现删失, 于是产生了双区间删失数据. 在双区间删失数据的研究中, 后续时间仅考虑发生右删失或区间删失的研究很多, 考虑右删失和区间删失同时存在的研究成果相对较少; 此外研究方法大多基于 Cox 模型. 本文在后续时间同时存在右删失和区间删失的这类双区间删失数据下建立比例优势模型, 利用 Stochastic EM 算法处理双区间删失数据并进行极大似然估计. 通过模拟研究评估了所提方法在有限样本下的优良性, 接着利用该方法分析了 AIDS 数据.

     

    Abstract: The incubation period is one of the important indicators of epidemiology and disease progression research, which plays an important role in disease prevention, control and treatment. The incubation period is the gap time between the virus infection and the manifestation of symptoms, and both occurrence times may be censored, resulting in doubly interval censored data. In the study of doubly interval censored data, there are many studies that only consider the occurrence of right censoring or interval censoring in the subsequent time, and there are relatively few studies that consider the simultaneous existence of right censoring and interval censoring. In addition, most of the research methods are based on the Cox model. In this paper, a proportional odds model is established under the doubly interval censored data with both right censored and interval censored in the subsequent time, and the Stochastic EM algorithm is used to process the doubly interval censored data and perform maximum likelihood estimation. The performance of the proposed method under finite samples is evaluated through simulation studies, and then the AIDS data is analyzed by this method.

     

/

返回文章
返回