Unearthing Valuable Macro Level Insights from Medical Data Using Time Series Methodologies

Abstract

This study applies time series methodologies, such as the Autoregressive Conditional Heteroscedasticity (ARCH) or Vector Error Correction (VEC) models to medical data for unearthing of valuable insights into trends in medical welfare of a population. Using data from the Centers for Disease Control and Prevention (CDC), study findings show three out of the four fastest growing cancers in the United States are cancers of the digestive system. These cancers are not the glamorously discussed cancers, such as lung cancer, or breast cancer. These are thyroid cancer, liver cancer, and pancreas cancer, cancers of the digestive system. ARCH models show the fastest growing cancer, thyroid cancer possesses characteristics of a progressive, yet randomly occurring series, a finding that ought to be worrying to policy makers, medical personnel, and researchers in realm of medicine. Application of VEC models to the data show incidence of thyroid cancer increases probability of each of lung cancer, and cancer of the uterus. In absence of a cointegration relation, this causal relation is shown to be unidirectional, with outcome risks of lung cancer cannot be dissociated from the digestive system. Study findings reveal cancers that are glamorized in the media have some of the lowest growth rates of all cancers, and perhaps are not the cancers deserving of the most attention. Regardless of demonstrated importance of micro level control studies for improvements to health outcomes, study findings provide evidence that studies of macro level data can yield valuable insights for policy initiatives, and directionality of micro level studies.