Beyond Volume: The Impact of Complex Healthcare Data on the Machine Learning Pipeline

摘要: From medical charts to national census, healthcare has traditionally operated under a paper-based paradigm. However, the past decade marked long and arduous transformation bringing into digital age. Ranging from electronic health records, digitized imaging laboratory reports, public datasets, today, now generates an incredible amount of information. Such wealth data presents exciting opportunity for integrated machine learning solutions address problems across multiple facets practice administration. Unfortunately, ability derive accurate informative insights requires more than execute models. Rather, deeper understanding on which models are run is imperative their success. While significant effort been undertaken develop able process volume obtained during analysis millions digitalized patient it important remember that represents only one aspect data. In fact, drawing increasingly diverse set sources, incredibly complex attributes must be accounted throughout pipeline. This chapter focuses highlighting such challenges, broken down three distinct components, each representing phase We begin with preprocessing, then move considerations model building, end challenges interpretation output. For component, we present discussion around as relates domain offer insight may impose efficiency techniques.