Challenges in Modeling Drug Shortage Events in the Pharmaceutical Domain

Meeting demand for medicines translates into a very difficult manufacturing, supply chain and commercial optimization problem for every pharmaceutical company. Time series data reflecting missing stock, disease incidence, sales, marketing and others can help with better demand forecasting and supply planning. Accurate demand forecast ensures an optimal functioning of the healthcare environment, however modeling multivariate data in this domain remains challenging. In this article, we describe the typical data types and the modeling problems faced by the pharmaceutical domain regarding missing stock forecasting. We want to motivate the research community to turn their attention onto a set of problems that are highly complex, under explored, but at the same time critically relevant for a highly important domain.