Skip to main page content
U.S. flag

An official website of the United States government

Dot gov

The .gov means it’s official.
Federal government websites often end in .gov or .mil. Before sharing sensitive information, make sure you’re on a federal government site.

Https

The site is secure.
The https:// ensures that you are connecting to the official website and that any information you provide is encrypted and transmitted securely.

Access keys NCBI Homepage MyNCBI Homepage Main Content Main Navigation
. 2020 Nov 12;17(22):8386.
doi: 10.3390/ijerph17228386.

Machine Learning for Mortality Analysis in Patients with COVID-19

Manuel Sánchez-Montañés  1 , Pablo Rodríguez-Belenguer  2 , Antonio J Serrano-López  2 , Emilio Soria-Olivas  2 , Yasser Alakhdar-Mohmara  3
Affiliations

Machine Learning for Mortality Analysis in Patients with COVID-19

Manuel Sánchez-Montañés et al. Int J Environ Res Public Health. .

Abstract

This paper analyzes a sample of patients hospitalized with COVID-19 in the region of Madrid (Spain). Survival analysis, logistic regression, and machine learning techniques (both supervised and unsupervised) are applied to carry out the analysis where the endpoint variable is the reason for hospital discharge (home or deceased). The different methods applied show the importance of variables such as age, O2 saturation at Emergency Rooms (ER), and whether the patient comes from a nursing home. In addition, biclustering is used to globally analyze the patient-drug dataset, extracting segments of patients. We highlight the validity of the classifiers developed to predict the mortality, reaching an appreciable accuracy. Finally, interpretable decision rules for estimating the risk of mortality of patients can be obtained from the decision tree, which can be crucial in the prioritization of medical care and resources.

Keywords: COVID-19; feature importance; graphical models; machine learning; survival analysis.

PubMed Disclaimer

Conflict of interest statement

The authors declare no conflict of interest.

Figures

Figure 1
Figure 1
Survival probability over time.
Figure 2
Figure 2
Survival analysis: (Left) Age. (Right) O2 saturation.
Figure 3
Figure 3
Survival analysis: (Left) Residential institution. (Right) heart rate.
Figure 4
Figure 4
(Top) Receiver Operating Curve (ROC) of the logistic regression model in training (left) and testing (right). (Bottom) Precision-Recall Curve (PRC) in training (left) and testing (right).
Figure 5
Figure 5
(Top) Receiver Operating Curve (ROC) for the decision tree model in training (left) and testing (right). (Bottom) Precision-Recall Curve (PRC) in training (left) and testing (right).
Figure 6
Figure 6
Decision tree.
Figure 7
Figure 7
Importance of the variables according to the random forest model.
Figure 8
Figure 8
(Top) Receiver Operating Curve (ROC) of the random forest model in training (left) and testing (right). (Bottom) Precision-Recall Curve (PRC) in training (left) and testing (right).
Figure 9
Figure 9
Tree augmented naive Bayes. The age was discretized considering as the cut-off values of 60 and 75 years and of 92% and 95% for O2 saturation.
Figure 10
Figure 10
(Top) Receiver Operating Curve (ROC) for the TAN model in the training (left) and testing (right) sets. (Bottom) Precision-Recall Curve (PRC) in the training (left) and testing (right) sets.
Figure 11
Figure 11
Receiver Operating Curves (ROCs) of the models. (Left) Training set. (Right) Testing set.
Figure 12
Figure 12
Precision-recall curve of the models. (Left) Training set. (Right) Testing set.
Figure 13
Figure 13
(Left) Original patient-drug matrix. (Right) Biclusters found by the co-clustering algorithm.
Figure 14
Figure 14
Representative drugs of each bicluster and their percentage of use for patients belonging to the bicluster versus the percentage for patients not in the bicluster.
See this image and copyright information in PMC

References

    1. Coronavirus Disease (COVID-19) Weekly Epidemiological Update and Weekly Operational Update. [(accessed on 31 August 2020)];2020 Available online: https://www.who.int/docs/default-source/coronaviruse/situation-reports/2....
    1. Condes E., Arribas J.R. Impact of COVID-19 on Madrid hospital system. Enferm. Infecc. Microbiol. Clin. 2020:11–21. doi: 10.1016/j.eimc.2020.06.005. - DOI - PMC - PubMed
    1. Wendel García P.D., Fumeaux T., Guercy P., Heuberger D.M., Montomoli J., Roche-Campo F., Schuepbach R.A., Hilty M.P., RISC-19-ICU Investigators Prognostic factors associated with mortality risk and disease progression in 639 critically ill patients with COVID-19 in Europe: Initial report of the international RISC-19-ICU prospective observational cohort. EClinicalMedicine. 2020;25:100449. doi: 10.1016/j.eclinm.2020.100449. - DOI - PMC - PubMed
    1. Berenguer J., Ryan P., Rodríguez-Baño J., Jarrín I., Carratalà J., Pachón J., Yllescas M., Arribas J. Characteristics and predictors of death among 4035 consecutively hospitalized patients with COVID-19 in Spain. Clin. Microbiol. Infect. 2020;29:82–97. doi: 10.1016/j.cmi.2020.07.024. - DOI - PMC - PubMed
    1. Rodríguez-Baño J., Pachón J., Carratalà J., Ryan P., Jarrín I., Yllescas M., Arribas J.R., Berenguer J. Treatment with tocilizumab or corticosteroids for COVID-19 patients with hyperinflammatory state: A multicentre cohort study (SAM-COVID-19) Clin. Microbiol. Infect. 2020 doi: 10.1016/j.cmi.2020.08.010. - DOI - PMC - PubMed

Publication types

LinkOut - more resources