Respiratory Development: Thinking Outside the Box – Big Data for Respiratory Medicines ?
1GlaxoSmithKline, 5 Crescent Drive, Philadelphia, PA, 19112, USA
2University of Kentucky, 111 Washington Avenue, Lexington, KY, USA
The concept of big data, which is typically characterized by volume, variety, veracity, and velocity has the capacity to transform the scientific approach to respiratory disease. This transformation ranges from a better understanding of the public health or population health approach to disease, to a drug safety and drug development, to alternative approaches to monitoring outbreaks and epidemics. Medical big data can generally be thought of as being in one of three classes: large numbers (often millions) of people with small numbers of parameters (such as mortality data or other administrative data); smaller numbers of people with large amounts of data ( such as micro array or genetic data); and, more recently large numbers of people with large amounts of data. Each of these classes present their own challenges and opportunities. For example, administrative data is often complicated by issues such as missing or incorrect data and potential biases, such as residual confounding or reverse causality. In addition, big data may be useful for hypothesis generation but is not rally able to test causality. Big data approaches have been used in respiratory in several applications: identification of asthma mortality patterns across different countries over many years (which pointed to certain classes of medications as being responsible); determining the relation between area of residence and respiratory mortality (and those changes over time); and identification of the relation between air pollution exposure and mortality. Future applications may help to define better targets for respiratory therapy development.