Personalized Medicine with Advanced Analytics

Practice of modern medicine demands personalized medicine (PM) to improve both quality of care and efficiency of the healthcare system. This is especially true as continued innovation in science and technology are delivering multiple interventions with different mechanism of actions for related comorbidity conditions. Real-world data (RWD) which typically includes a large, diverse, and heterogeneous patient population presents unique opportunity for PM research. In addition to natural history of disease and comparative effectiveness research, personalized medicine, that is who should initiate what and when, has been another central theme facing policy makers, physicians and patients. In the meantime, real-world studies (RWSs) usually come with high-dimensional and different types of covariates, relatively high proportion of missingness, more intercurrent events, a necessity to address confounding and quantify potential complex relationship between intervention and outcome. Therefore, advanced analytics are usually more suitable when addressing personalized medicine research questions. A few active areas include predictive modeling, natural language processing for free text to augment structured data, empirical subgroup identification, individualized treatment regime, and dynamic treatment regimens to define sequence of individually tailored decision rules. This chapter will review the latest developments in advanced analytics with the objective of facilitating adoption of personalized medicine and a focus of their relevance in generating robust real-world evidence (RWE).