Mortality Prediction from Hospital-Acquired Infections in Trauma Patients Using an Unbalanced Dataset
1
조회
0
다운로드
0
인용
0
참고문헌
키워드
초록
Objectives: Machine learning has been widely used to predict diseases, and it is used to derive impressive knowledge in thehealthcare domain. Our objective was to predict in-hospital mortality from hospital-acquired infections in trauma patientson an unbalanced dataset. Methods: Our study was a cross-sectional analysis on trauma patients with hospital-acquired infectionswho were admitted to Shiraz Trauma Hospital from March 20, 2017, to March 21, 2018. The study data was obtainedfrom the surveillance hospital infection database. The data included sex, age, mechanism of injury, body region injured, severityscore, type of intervention, infection day after admission, and microorganism causes of infections. We developed ourmortality prediction model by random under-sampling, random over-sampling, clustering (k-mean)-C5.0, SMOTE-C5.0,ADASYN-C5.5, SMOTE-SVM, ADASYN-SVM, SMOTE-ANN, and ADASYN-ANN among hospital-acquired infectionsin trauma patients. All mortality predictions were conducted by IBM SPSS Modeler 18. Results: We studied 549 individualswith hospital-acquired infections in a trauma hospital in Shiraz during 2017 and 2018. Prediction accuracy before balancingof the dataset was 86.16%. In contrast, the prediction accuracy for the balanced dataset achieved by random under-sampling,random over-sampling, clustering (k-mean)-C5.0, SMOTE-C5.0, ADASYN-C5.5, and SMOTE-SVM was 70.69%, 94.74%,93.02%, 93.66%, 90.93%, and 100%, respectively. Conclusions: Our findings demonstrate that cleaning an unbalanced datasetincreases the accuracy of the classification model. Also, predicting mortality by a clustered under-sampling approach wasmore precise in comparison to random under-sampling and random over-sampling methods.
Abstract
Objectives: Machine learning has been widely used to predict diseases, and it is used to derive impressive knowledge in thehealthcare domain. Our objective was to predict in-hospital mortality from hospital-acquired infections in trauma patientson an unbalanced dataset. Methods: Our study was a cross-sectional analysis on trauma patients with hospital-acquired infectionswho were admitted to Shiraz Trauma Hospital from March 20, 2017, to March 21, 2018. The study data was obtainedfrom the surveillance hospital infection database. The data included sex, age, mechanism of injury, body region injured, severityscore, type of intervention, infection day after admission, and microorganism causes of infections. We developed ourmortality prediction model by random under-sampling, random over-sampling, clustering (k-mean)-C5.0, SMOTE-C5.0,ADASYN-C5.5, SMOTE-SVM, ADASYN-SVM, SMOTE-ANN, and ADASYN-ANN among hospital-acquired infectionsin trauma patients. All mortality predictions were conducted by IBM SPSS Modeler 18. Results: We studied 549 individualswith hospital-acquired infections in a trauma hospital in Shiraz during 2017 and 2018. Prediction accuracy before balancingof the dataset was 86.16%. In contrast, the prediction accuracy for the balanced dataset achieved by random under-sampling,random over-sampling, clustering (k-mean)-C5.0, SMOTE-C5.0, ADASYN-C5.5, and SMOTE-SVM was 70.69%, 94.74%,93.02%, 93.66%, 90.93%, and 100%, respectively. Conclusions: Our findings demonstrate that cleaning an unbalanced datasetincreases the accuracy of the classification model. Also, predicting mortality by a clustered under-sampling approach wasmore precise in comparison to random under-sampling and random over-sampling methods.
참고문헌 (26)
- Lee DG, "Discovering medical knowledge using association rule mining in young adults with acute myocardial infarction", J Med Syst, 37(2), pp.9896~ (2013)
- Guo H, "Learning from class-imbalanced data : review of methods and applications", Expert Syst Appl, 73, pp.220~239 (2017)
- Li Y, "Adapted ensemble classification algorithm based on multiple classifier system and feature selection for classifying multi-class imbalanced data", Knowl Based Syst, 94, pp.88~104 (2016)
- Krawczyk B, "Learning from imbalanced data : open challenges and future directions", Prog Artif Intell, 5(4), pp.221~232 (2016)
- Wallace WC, "Nosocomial infections in the surgical intensive care unit : a difference between trauma and surgical patients", Am Surg, 65(10), pp.987~990 (1999)
- Burke JP, "Infection control : a problem for patient safety", N Engl J Med, 348(7), pp.651~656 (2003)
- Anderson RN, "Deaths: leading causes for 1999" (2001), National Center for Health Statistics
- Czaja AS, "Late outcomes of trauma patients with infections during index hospitalization", J Trauma, 67(4), pp.805~814 (2009)
- Glance LG, "Increases in mortality, length of stay, and cost associated with hospital-acquired infections in trauma patients", Arch Surg, 146(7), pp.794~801 (2011)
- Sheng WH, "Risk factors affecting in-hospital mortality in patients with nosocomial infections", J Formos Med Assoc, 106(2), pp.110~118 (2007)
- Yadollahi M, "The accuracy of GAP and MGAP scoring systems in predicting mortality in trauma : a diagnostic accuracy study", Adv J Emerg Med, 4(3), pp.e73~ (2020)
- Spelmen VS, "A review on handling imbalanced data", Proceedings of 2018 International Conference on Current Trends towards Converging Technologies (ICCTCT), pp.1~11 (2018)
- Saarela M, "Predicting hospital associated disability from imbalanced data using supervised learning", Artif Intell Med, 95, pp.88~95 (2019)
- Klikowski J, "Multi sampling random subspace ensemble for imbalanced data stream classification", Progress in computer recognition systems, pp.360~369 (2019), Springer
- Roumani YF, "Classifying highly imbalanced ICU data", Health Care Manag Sci, 16(2), pp.119~128 (2013)
- Paoin W, "Lessons learned from data mining of WHO mortality database", Methods Inf Med, 50(4), pp.380~385 (2011)
- Wirth R, "CRISP-DM: towards a standard process model for data mining", Proceedings of the 4th International Conference on the Practical Applications of Knowledge Discovery and Data Mining, pp.29~39 (2000)
- Bolandparvaz S, "Injury patterns among various age and gender groups of trauma patients in southern Iran: a cross-sectional study", Medicine (Baltimore), 96(41), pp.e7812~ (2017)
- Alonso SG, "Data mining algorithms and techniques in mental health: a systematic review", J Med Syst, 42(9), pp.161~ (2018)
- Lin CL, "Evaluation of CART, CHAID, and QUEST algorithms : a case study of construction defects in Taiwan", J Asian Archit Build Eng, 18(6), pp.539~553 (2019)
- Chawla NV, "SMOTE : synthetic minority over-sampling technique", J Artif Intell Res, 16, pp.321~357 (2002)
- Arisholm E, "A systematic and comprehensive investigation of methods to build and evaluate fault prediction models", J Syst Softw, 83(1), pp.2~17 (2010)
- Yen SJ, "Cluster-based under-sampling approaches for imbalanced data distributions", Expert Syst Appl, 36(3), pp.5718~5727 (2009)
- Rahman MM, "Cluster based under-sampling for unbalanced cardiovascular data", Proceedings of the World Congress on Engineering (WCE) (2013)
- Onan A, "Consensus clustering-based undersampling approach to imbalanced learning", Sci Program, 2019, pp.5901087~ (2019)
- Tyagi AK, "Performance analysis of undersampling and over-sampling techniques for solving class imbalance problem", Proceedings of International Conference on Sustainable Computing in Science, Technology and Management (SUSCOM) (2019)