Cancer therapeutics have seen many success stories over the past few decades. However while some forms of cancer respond very well to treatment, a sizeable majority of cancers are not that responsive, contributing to cancer being a common cause of death in the United States, second only to cardiovascular diseases. An important contributing factor behind this is that every cancer is unique to a person, hence two cancers in two different people may be of a very different nature (or type) even if have superficial similarities. Due to the lack of an adequate understanding of how cancer develops, we do not have treatments that can reliably work across different cancer types.
In such cases, an often-used approach, together with longer-term efforts to understand the genetic or immunological factors behind specific types of cancer, is the usage of combination therapy where different drugs are administered either at the same time or one after the other, and the drugs act synergistically.
Our project focuses on enabling a better understanding of the potential interactions and side effects of different combination-therapeutic approaches. Our goal is to implement a platform which enables effective analysis and visualization of related treatment/treatment-outcome data. Since there are privacy issues associated with the acquisition of such a dataset, we focus our efforts on publicly available data on multiple causes of death (MCD). The MCD dataset has information about existing conditions at the time of death in all recorded mortality cases in the United States, and allows us to study the effects of different combinations of factors (including drug related side effects) on mortality.
In this talk, several intuitive and counter-intuitive combinations of common and not-so-common conditions found in the MCD dataset will be highlighted through queries and visualizations performed on ourplatform and their implications for improving combination therapy will be discussed. As will be pointed out, one can use our platform not only to study combinations of conditions, but also for testing general hypotheses about causes of mortality. Some algorithmic issues associated with addressing the queries at scale will also be discussed. Our platform offers an intuitive, systematic and objective summary of the publicly available MCD dataset and is open for anyone to use for data exploration and hypothesis testing purposes.