Machine Learning Based Sentiment Text Classification for Evaluating Treatment Quality of Discharge Summary
Open Access
- 23 May 2020
- journal article
- research article
- Published by MDPI AG in Information
- Vol. 11 (5) , 281
Abstract
Patients’ discharge summaries (documents) are health sensors that are used for measuring the quality of treatment in medical centers. However, extracting information automatically from discharge summaries with unstructured natural language is considered challenging. These kinds of documents include various aspects of patient information that could be used to test the treatment quality for improving medical-related decisions. One of the significant techniques in literature for discharge summaries classification is feature extraction techniques from the domain of natural language processing on text data. We propose a novel sentiment analysis method for discharge summaries classification that relies on vector space models, statistical methods, association rule, and extreme learning machine autoencoder (ELM-AE). Our novel hybrid model is based on statistical methods that build the lexicon in a domain related to health and medical records. Meanwhile, our method examines treatment quality based on an idea inspired by sentiment analysis. Experiments prove that our proposed method obtains a higher F1 value of 0.89 with good TPR (True Positive Rate) and FPR (False Positive Rate) values compared with various well-known state-of-the-art methods with different size of training and testing datasets. The results also prove that our method provides a flexible and effective technique to examine treatment quality based on positive, negative, and neutral terms for sentence-level in each discharge summary.Keywords
Funding Information
- National Natural Science Foundation of China (U1811262)
This publication has 38 references indexed in Scilit:
- Assessing the impact of the introduction of an electronic hospital discharge system on the completeness and timeliness of discharge communication: a before and after studyBMC Health Services Research, 2017
- Deep Learning of Semisupervised Process Data With Hierarchical Extreme Learning Machine and Soft Sensor ApplicationIEEE Transactions on Industrial Electronics, 2017
- Improving communication of medication changes using a pharmacist-prepared discharge medication management summaryInternational Journal of Clinical Pharmacy, 2017
- Generalized extreme learning machine autoencoder and a new deep neural networkNeurocomputing, 2017
- Predicting early psychiatric readmission with natural language processing of narrative discharge summariesTranslational Psychiatry, 2016
- SentiTurkNet: a Turkish polarity lexicon for sentiment analysisLanguage Resources and Evaluation, 2015
- Supporting information retrieval from electronic health records: A report of University of Michigan’s nine-year experience in developing and using the Electronic Medical Record Search Engine (EMERSE)Journal of Biomedical Informatics, 2015
- Deep learningNature, 2015
- Trends in extreme learning machines: A reviewNeural Networks, 2015
- Extreme learning machine: Theory and applicationsNeurocomputing, 2006