A robust network topology approach for survival analysis

Network topology can be used to simplify the complexity of the datasets.  We are exploring its function in performing survival analysis to identify the most important factors that contributed to the survival time from diagnosis to death.  This technique has the potential to illustrate easily some types of complex interactions in dataset.  Then, based on those interactions, the most important factors in survival analysis are identified.  However, network topology based on classical estimator is extremely sensitive to outlying observations, i.e., conclusions drawn from contaminated network topology could be misleading.  Hence, in this paper, the classical estimator, i.e., classical correlation matrix of network topology is substituted with robust estimator, i.e., robust correlation matrix which is developed based on Index Set Equality (ISE).  Then, the interpretation of that robust network topology is delivered by using centrality measure, i.e., degree centrality.  A case study of the survival time for cervical cancer patients is presented and discussed. Robust network topology revealed that the most important factors that influence the survival of cervical cancer patients is stage at diagnosis (STG).  The higher stage of cervical cancer led to shorter survival time of cervical cancer patients.  Consequently, early diagnosis is very important.  Early diagnosis of cancer allows early intervention to try to slow or prevent cancer development and lethality, hence, the survival improves.