Copula-based Bivariate Model for Association between Lung Disease and Skin Lesion

Abstract

Chronic lung disease has been reported to occur from chronic arsenic exposure in several studies. Skin lesion has been considered as a hallmarked sign of arsenic exposure. Several studies also found the elevated prevalence of lung disease in presence of skin lesion. The present study examines the association of lung disease with skin lesion considering lung disease and skin lesion as binary and ordinal observations respectively. A bivariate limited dependent model is an appropriate statistical tool for this type of modeling. This paper develops a bivariate model specifying marginal regression models for the responses, and combines them via a copula to form a joint model. The commonly used Gaussian copula has been used to model the association. The maximum likelihood (ML) method has been used to estimate the model. The application shows a significant positive association between lung disease and skin lesion.