Applications of Clinical Text Mining

Abstract

This chapter will present the state of the art for a number of applications of clinical text mining such as detection and prediction of healthcare associated infections (HAI), detection of adverse drug events (ADE), followed by resources for adverse drug event detection and continuing with an application of automatic assignment and validation of ICD-10 diagnosis codes. An application of automatic mapping of ICD-10 diagnosis codes to SNOMED CT will also be presented. This chapter then continues with automatic summarisation of patient record, simplification of patient records text and natural language generation of patient record text. Techniques for searching and retrieving patients from patient records for cohort studies are described, and finally some classic systems in medical decision support such as MYCIN are reviewed. Finally an overview of IBM Watson Health is provided. 10.1 Detection and Prediction of Healthcare Associated Infections (HAIs) This section explains what a healthcare associated infection (HAI) is, why it is important to monitor and predict them and how to do it automatically using the information from the electronic patient records. The performance of systems for detecting HAIs will be compared. This section will also describe in what extent and where such surveillance systems for detecting HAIs are used in practice and commercially. 10.1.1 Healthcare Associated Infections (HAIs) Healthcare associated infections (HAIs) are plaguing healthcare with suffering patients and heavy costs for society. Healthcare associated infections are also called hospital associated infections or nosocomial infections.