Publicação
Reduction of emergency department returns after discharge from hospital: Machine learning model to predict emergency department returns 30 days post hospital discharge for medical patients
| dc.contributor.advisor | Henriques, Roberto André Pereira | |
| dc.contributor.advisor | Lopes, Sílvia | |
| dc.contributor.author | Bouziri, Ikram | |
| dc.date.accessioned | 2023-02-10T14:42:10Z | |
| dc.date.available | 2023-02-10T14:42:10Z | |
| dc.date.issued | 2023-01-24 | |
| dc.description | Dissertation presented as the partial requirement for obtaining a Master's degree in Data Science and Advanced Analytics, specialization in Business Analytics | pt_PT |
| dc.description.abstract | Post-hospital discharge returns to emergency departments are associated with reducing the efficiency of the emergency department (ED) utilisation and the quality of healthcare. These returns are often related to the nature of the disease and/or inadequate care. This thesis aims to develop a machine-learning model that predicts ED returns within 30 days of inpatient discharge from Portuguese public hospitals. Different binary classification models were trained and evaluated with a particular focus on sensitivity (predictive power of the critical class of returning patients). The selected model was the Extreme gradient boost Classifier, which showed the best performance on recall and the other considered performance metrics. A cohort of 93 449 medical hospitalisations of adult patients discharged between January 1st, 2018, and December 31st, 2019, was assembled with diagnoses details to be used in this study. According to the problem's requirement, the recall was the performance metric to be maximised. Therefore, Performance optimisation methods were considered, and the final model resulted in a recall of 84.38%, precision of 84.35%, F1 score of 84.36% and accuracy of 84.10%. Future deployment and integration of this ED return predictive analytics into the inpatient care workflow may allow identifying patients that require targeted care interventions that reduce overall healthcare expense and improve health outcomes. | pt_PT |
| dc.identifier.tid | 203220420 | pt_PT |
| dc.identifier.uri | http://hdl.handle.net/10362/148993 | |
| dc.language.iso | eng | pt_PT |
| dc.rights.uri | http://creativecommons.org/licenses/by/4.0/ | pt_PT |
| dc.subject | Machine Learning | pt_PT |
| dc.subject | Binary Classification | pt_PT |
| dc.subject | Emergency Department Returns | pt_PT |
| dc.subject | Performance Metrics | pt_PT |
| dc.subject | Performance Optimisation | pt_PT |
| dc.title | Reduction of emergency department returns after discharge from hospital: Machine learning model to predict emergency department returns 30 days post hospital discharge for medical patients | pt_PT |
| dc.type | master thesis | |
| dspace.entity.type | Publication | |
| rcaap.rights | openAccess | pt_PT |
| rcaap.type | masterThesis | pt_PT |
| thesis.degree.name | Mestrado em Ciência de Dados e Métodos Analíticos Avançados, especialização em Métodos Analíticos para a Gestão | pt_PT |