2018
Prata, M.; Peixoto, H.; MacHado, J.; Abelha, A.
Data mining in urgency department: Medical specialty discharge prediction Proceedings Article
Em: J., Guerra H. Gomes L. M. Machado (Ed.): pp. 28-35, EUROSIS, 2018, ISBN: 9789492859037, (cited By 2; Conference of 16th International Industrial Simulation Conference, ISC 2018 ; Conference Date: 6 June 2018 Through 8 June 2018; Conference Code:137024).
Resumo | Links | BibTeX | Etiquetas: Artificial intelligence; Data mining; Feature extraction; Forecasting; Hospitals; Learning algorithms; Learning systems; Predictive analytics, Bagging algorithms; Classification accuracy; Cross validation; Discharge predictions; F1 scores; Medical specialties; Quality of data; Urgency Department, Medical computing
@inproceedings{Prata201828,
title = {Data mining in urgency department: Medical specialty discharge prediction},
author = {M. Prata and H. Peixoto and J. MacHado and A. Abelha},
editor = {Guerra H. Gomes L.M. Machado J.},
url = {https://www.scopus.com/inward/record.uri?eid=2-s2.0-85049981921&partnerID=40&md5=3cc8bef2ee13f2fbe0c31b22754c6e21},
isbn = {9789492859037},
year = {2018},
date = {2018-01-01},
journal = {16th International Industrial Simulation Conference 2018, ISC 2018},
pages = {28-35},
publisher = {EUROSIS},
abstract = {The aim of this paper is to analyze and process a dataset to predict the Medical Specialty (MS) discharge in a hospital Urgency Department (UD). MS discharge is the medical specialty in which a patient gets discharged from the UD. This predictive analysis would improve medical and staff performance, not to mention, less time consuming to the patient and less expensive to both patient and the hospital. However, it is a challenging task due to the quality of data retrieved from UDs that's usually non-treated, with a lot of irrelevant information, sparse and, sometimes, incomplete. This predictive analysis is obtained through Data Mining techniques and machine learning algorithms in Weka environment. It was concluded that feature selection and structured modelling are important factors that affect classification accuracy. It was also concluded as that randomly decreasing or increasing dataset information by varying patient values does not assist directly in increasing accuracy for the prediction of MS discharge. The best results were achieved using the Bagging algorithm with a REPTree classifier and a ten-fold cross-validation, achieving 91.96 % of accuracy and 0.85 F1-score. © 2018 EUROSIS. All rights reserved.},
note = {cited By 2; Conference of 16th International Industrial Simulation Conference, ISC 2018 ; Conference Date: 6 June 2018 Through 8 June 2018; Conference Code:137024},
keywords = {Artificial intelligence; Data mining; Feature extraction; Forecasting; Hospitals; Learning algorithms; Learning systems; Predictive analytics, Bagging algorithms; Classification accuracy; Cross validation; Discharge predictions; F1 scores; Medical specialties; Quality of data; Urgency Department, Medical computing},
pubstate = {published},
tppubtype = {inproceedings}
}
The aim of this paper is to analyze and process a dataset to predict the Medical Specialty (MS) discharge in a hospital Urgency Department (UD). MS discharge is the medical specialty in which a patient gets discharged from the UD. This predictive analysis would improve medical and staff performance, not to mention, less time consuming to the patient and less expensive to both patient and the hospital. However, it is a challenging task due to the quality of data retrieved from UDs that's usually non-treated, with a lot of irrelevant information, sparse and, sometimes, incomplete. This predictive analysis is obtained through Data Mining techniques and machine learning algorithms in Weka environment. It was concluded that feature selection and structured modelling are important factors that affect classification accuracy. It was also concluded as that randomly decreasing or increasing dataset information by varying patient values does not assist directly in increasing accuracy for the prediction of MS discharge. The best results were achieved using the Bagging algorithm with a REPTree classifier and a ten-fold cross-validation, achieving 91.96 % of accuracy and 0.85 F1-score. © 2018 EUROSIS. All rights reserved.