This talk will discuss the increasing role of data science and machine learning in clinical research and development. Emphasis will be placed on the need to work closely with domain science experts to clearly scope and define research questions. A case-study will be presented based on a detailed exploratory analysis of large psoriatic arthritis program, where hundreds of measurements were collected on patients over many time points. From a technical perspective we will examine solutions based multivariate longitudinal mixture models that handle mixed data types.
Drawing on the experience of how Bayesian methods moved into practice in drug development, we will also discuss how data science and machine learning can achieve greater impact in this setting. The discussion will focus on the need for incorporating statistical principles to provide compelling solutions to a wider set of scientific stakeholders
|