The Role of Big Data in Personalized Medicine

Abstract

Standard of care is defined by the therapeutic intervention that significantly improved the outcome of a cohort of patients, ideally in a randomized trial. It means that for patients on the fringe of the general population, the standard of care could actually not be the best treatment. In order to better personalize treatments, new approaches are required. The digitalization of medicine has generated large databases that can be explored in order to define comprehensive patient phenotypes and better adapt treatments to each profile. Big Data is defined by data that is too large to be analyzed with traditional processing software. They are typically of high volume, variety, and veracity, and are generated with high velocity. In the first part of this chapter, we will define what Big Data is and explore the sources of data available: which medical data can be harvested and used; how granular and structured the data should be; what is an ontology and why they are essential for data mining; what is a clinical data warehouse (CDW); and what are the main platforms available for research? In the second part of the chapter, we will describe the methods used to explore large datasets, with a focus on machine learning, deep learning, and natural language processing. In the third and last part of the chapter, the challenges in Big Data analytics will be discussed: how to implement common governance and a data-sharing culture? How to appropriately protect the data? Several ethical issues will also be raised: if therapeutic decisions are entirely derived from existing data, what are the implications for the practice and evolution of medicine?

Ähnliche Arbeiten