Automated Mortality Prediction in Critically-ill Patients with Thrombosis using Machine Learning
| Autor: | Vassiliki Danilatou, Vasilios Katos, Theodoros Kostoulas, Christos Tzagkarakis, Alexandros Kanterakis, Despoina Antonakaki |
|---|---|
| Jazyk: | angličtina |
| Rok vydání: | 2020 |
| Předmět: |
Critically ill
business.industry Mortality rate medicine.disease Machine learning computer.software_genre Thrombosis Icu admission 03 medical and health sciences 0302 clinical medicine 030228 respiratory system Intensive care medicine 030212 general & internal medicine Artificial intelligence Mortality prediction Medical assessment business Venous thromboembolism computer |
| Zdroj: | 20th International Conference on BioInformatics and BioEngineering BIBE |
| Popis: | Venous thromboembolism (VTE) is the third most\ud common cardiovascular condition. Some high risk patients diagnosed with VTE need immediate treatment and monitoring\ud in intensive care units (ICU) as the mortality rate is high.\ud Most of the published predictive models for ICU mortality give\ud information on in-hospital mortality using data recorded in the\ud first day of ICU admission. The purpose of the current study is to\ud predict in-hospital and after-discharge mortality in patients with\ud VTE admitted to ICU using a machine learning (ML) framework.\ud We studied 2,468 patients from the Medical Information Mart\ud for Intensive Care (MIMIC-III) database, admitted to ICU with\ud a diagnosis of VTE. We formed ML classification tasks for\ud early and late mortality prediction. In total, 1,471 features were\ud extracted for each patient, grouped in seven categories each\ud representing a different type of medical assessment. We used an\ud automated ML platform, JADBIO, as well as a class balancing\ud combined with a Random Forest classifier, in order to evaluate the\ud importance of class imbalance. Both methods showed significant\ud ability in prediction of early mortality (AUC=0.92). Nevertheless,\ud the task of predicting late mortality was less efficient (AUC=0.82).\ud To the best of our knowledge, this is the first study in which\ud ML is used to predict short-term and long-term mortality for\ud ICU patients with VTE based on a multitude of clinical features\ud collected over time. |
| Databáze: | OpenAIRE |
| Externí odkaz: |
|