Predicting Mortality Risk among Elderly Inpatients with Pneumonia: A Machine Learning Approach

Victor Silva, Damires Fernandes, Alex Rêgo

2022

Abstract

Community-acquired Pneumonia (CAP) is a serious respiratory infection that can cause life-threatening risk in people of different ages, especially in elderly inpatients. Regarding this age group, mortality rates by CAP still can reach 30% of all respiratory causes of death. In this work, we propose a machine learning approach to predict mortality risk among elderly inpatients with CAP. The approach uses real world data of elderly people with CAP from a hospital in Brazil, collected from 2018 to 2021. Based on patients data as learning features, our approach is able not only to classify patients at risk of mortality during hospitalization, but also to estimate the probability concerning the prediction. Some classification models have been examined and, among them, the best performance in terms of Area under ROC Curve (AUC) value has been achieved by the Logistic Regression (LR) classifier (AUC=0.81). Accomplished results show that the presented approach outperforms CURB-65 score as baseline in terms of both AUC values and probability of patient death. Besides, our approach is able to output probabilities ranging from 50 to 99% w.r.t. positive classification, i.e., patients that may come to death. A statistical test confirms that the presented approach outperforms the baseline provided by the CURB-65.

Paper Citation

in Harvard Style

Silva V., Fernandes D. and Rêgo A. (2022). Predicting Mortality Risk among Elderly Inpatients with Pneumonia: A Machine Learning Approach. In Proceedings of the 24th International Conference on Enterprise Information Systems - Volume 2: ICEIS, ISBN 978-989-758-569-2, pages 344-354. DOI: 10.5220/0011043300003179

in Bibtex Style

@conference{iceis22,
author={Victor Silva and Damires Fernandes and Alex Rêgo},
title={Predicting Mortality Risk among Elderly Inpatients with Pneumonia: A Machine Learning Approach},
booktitle={Proceedings of the 24th International Conference on Enterprise Information Systems - Volume 2: ICEIS,},
year={2022},
pages={344-354},
publisher={SciTePress},
organization={INSTICC},
doi={10.5220/0011043300003179},
isbn={978-989-758-569-2},
}

in EndNote Style

TY - CONF

JO - Proceedings of the 24th International Conference on Enterprise Information Systems - Volume 2: ICEIS,
TI - Predicting Mortality Risk among Elderly Inpatients with Pneumonia: A Machine Learning Approach
SN - 978-989-758-569-2
AU - Silva V.
AU - Fernandes D.
AU - Rêgo A.
PY - 2022
SP - 344
EP - 354
DO - 10.5220/0011043300003179