A Natural Language Processing Framework for Assessing Hospital Readmissions for Patients With COPD

• Published in: IEEE journal of biomedical and health informatics Vol. 22; no. 2; pp. 588 - 596
• Main Authors: Agarwal, Ankur, Baechle, Christopher, Behara, Ravi, Zhu, Xingquan
• Format: Journal Article
• Language: English
• Published: United States IEEE 01.03.2018; The Institute of Electrical and Electronics Engineers, Inc. (IEEE)
• Subjects: Algorithms; Area Under Curve; Bayesian analysis; Chronic obstructive pulmonary disease; Computer applications; Data mining; Data processing; decision support systems; Decision Support Systems, Clinical; Diseases; feature extraction; Federal legislation; Hazards; Hospitals; Humans; Language; Learning algorithms; Legislation; Libraries; Lung diseases; Machine learning; Medical diagnosis; Medical diagnostic imaging; Medical Informatics - methods; medical information systems; Natural Language Processing; Obstructive lung disease; Patient admissions; Patient Readmission - statistics & numerical data; Pipelines; Pulmonary Disease, Chronic Obstructive - epidemiology; Pulmonary Disease, Chronic Obstructive - therapy; Statistical tests
• ISSN: 2168-2194; 2168-2208; 2168-2208
• DOI: 10.1109/JBHI.2017.2684121

With the passage of recent federal legislation, many medical institutions are now responsible for reaching target hospital readmission rates. Chronic diseases account for many hospital readmissions and chronic obstructive pulmonary disease has been recently added to the list of diseases for which the United States government penalizes hospitals incurring excessive readmissions. Though there have been efforts to statistically predict those most in danger of readmission, a few have focused primarily on unstructured clinical notes. We have proposed a framework, which uses natural language processing to analyze clinical notes and predict readmission. Many algorithms within the field of data mining and machine learning exist, so a framework for component selection is created to select the best components. Naïve Bayes using Chi-Squared feature selection offers an AUC of 0.690 while maintaining fast computational times.