Executive Summary : | Life-time data analysis has a long history, with a focus on biological and industrial areas. Recurrent events are a common phenomenon in survival time studies, particularly in public health. However, existing literature only assumes the time to event to be continuos, without considering discrete time to alternating recurrent events. This work aims to model recurrent pulmonary exacerbations, such as asthma attacks. The problem started when analyzing the survival experiences of a group of smokers treated at the Institute of Pulmocare and Research, Kolkata. The usual statistical methods on recurrent events were applied, but they were unable to provide satisfactory results due to unutilized additional information collected during patients' hospital visits. Most cases follow severe pulmonary exacerbations, with some less severe attacks in between. Data on the number of attacks until the severe attack occurred and after the severe attack, there are some less severe attacks, leading to partial cure for a considerable time.
The project aims to model the relationship between sub-attacks in successive cycles and refine the data collection scheme by properly documenting the number of sub-attacks before any severe pulmonary exacerbation and the number of sub-attacks before a substantial partial cure. This will require the introduction of techniques for discrete alternating recurrent events, where previous literatures are nonexistent. |