Comparing machine learning algorithms for predicting COVID-19 mortality.
Artificial intelligence
COVID-19
Coronavirus
Machine learning
Prediction hospital mortality
Journal
BMC medical informatics and decision making
ISSN: 1472-6947
Titre abrégé: BMC Med Inform Decis Mak
Pays: England
ID NLM: 101088682
Informations de publication
Date de publication:
04 01 2022
04 01 2022
Historique:
received:
26
05
2021
accepted:
28
12
2021
entrez:
5
1
2022
pubmed:
6
1
2022
medline:
12
1
2022
Statut:
epublish
Résumé
The coronavirus disease (COVID-19) hospitalized patients are always at risk of death. Machine learning (ML) algorithms can be used as a potential solution for predicting mortality in COVID-19 hospitalized patients. So, our study aimed to compare several ML algorithms to predict the COVID-19 mortality using the patient's data at the first time of admission and choose the best performing algorithm as a predictive tool for decision-making. In this study, after feature selection, based on the confirmed predictors, information about 1500 eligible patients (1386 survivors and 144 deaths) obtained from the registry of Ayatollah Taleghani Hospital, Abadan city, Iran, was extracted. Afterwards, several ML algorithms were trained to predict COVID-19 mortality. Finally, to assess the models' performance, the metrics derived from the confusion matrix were calculated. The study participants were 1500 patients; the number of men was found to be higher than that of women (836 vs. 664) and the median age was 57.25 years old (interquartile 18-100). After performing the feature selection, out of 38 features, dyspnea, ICU admission, and oxygen therapy were found as the top three predictors. Smoking, alanine aminotransferase, and platelet count were found to be the three lowest predictors of COVID-19 mortality. Experimental results demonstrated that random forest (RF) had better performance than other ML algorithms with accuracy, sensitivity, precision, specificity, and receiver operating characteristic (ROC) of 95.03%, 90.70%, 94.23%, 95.10%, and 99.02%, respectively. It was found that ML enables a reasonable level of accuracy in predicting the COVID-19 mortality. Therefore, ML-based predictive models, particularly the RF algorithm, potentially facilitate identifying the patients who are at high risk of mortality and inform proper interventions by the clinicians.
Sections du résumé
BACKGROUND
The coronavirus disease (COVID-19) hospitalized patients are always at risk of death. Machine learning (ML) algorithms can be used as a potential solution for predicting mortality in COVID-19 hospitalized patients. So, our study aimed to compare several ML algorithms to predict the COVID-19 mortality using the patient's data at the first time of admission and choose the best performing algorithm as a predictive tool for decision-making.
METHODS
In this study, after feature selection, based on the confirmed predictors, information about 1500 eligible patients (1386 survivors and 144 deaths) obtained from the registry of Ayatollah Taleghani Hospital, Abadan city, Iran, was extracted. Afterwards, several ML algorithms were trained to predict COVID-19 mortality. Finally, to assess the models' performance, the metrics derived from the confusion matrix were calculated.
RESULTS
The study participants were 1500 patients; the number of men was found to be higher than that of women (836 vs. 664) and the median age was 57.25 years old (interquartile 18-100). After performing the feature selection, out of 38 features, dyspnea, ICU admission, and oxygen therapy were found as the top three predictors. Smoking, alanine aminotransferase, and platelet count were found to be the three lowest predictors of COVID-19 mortality. Experimental results demonstrated that random forest (RF) had better performance than other ML algorithms with accuracy, sensitivity, precision, specificity, and receiver operating characteristic (ROC) of 95.03%, 90.70%, 94.23%, 95.10%, and 99.02%, respectively.
CONCLUSION
It was found that ML enables a reasonable level of accuracy in predicting the COVID-19 mortality. Therefore, ML-based predictive models, particularly the RF algorithm, potentially facilitate identifying the patients who are at high risk of mortality and inform proper interventions by the clinicians.
Identifiants
pubmed: 34983496
doi: 10.1186/s12911-021-01742-0
pii: 10.1186/s12911-021-01742-0
pmc: PMC8724649
doi:
Types de publication
Journal Article
Langues
eng
Sous-ensembles de citation
IM
Pagination
2Informations de copyright
© 2022. The Author(s).
Références
PeerJ. 2020 Sep 28;8:e10083
pubmed: 33062451
PLoS One. 2020 Oct 19;15(10):e0240711
pubmed: 33075088
Intern Emerg Med. 2020 Nov;15(8):1435-1443
pubmed: 32812204
PLoS One. 2020 Jul 30;15(7):e0236618
pubmed: 32730358
Mod Pathol. 2021 Mar;34(3):522-531
pubmed: 33067522
Biomed Eng Online. 2017 Nov 2;16(1):125
pubmed: 29096638
Avicenna J Med Biotechnol. 2013 Jul;5(3):148-57
pubmed: 23919118
J Med Internet Res. 2020 Jun 29;22(6):e19786
pubmed: 32540845
Bioinformatics. 2005 Aug 1;21(15):3301-7
pubmed: 15905277
Lancet Digit Health. 2020 Oct;2(10):e516-e525
pubmed: 32984797
Comput Methods Programs Biomed. 2019 Mar;170:23-29
pubmed: 30712601
Eur J Epidemiol. 2020 Aug;35(8):733-742
pubmed: 32780189
Chaos Solitons Fractals. 2020 Sep;138:110137
pubmed: 32834583
Clin Transl Sci. 2019 Sep;12(5):519-528
pubmed: 31112000
JMIR Med Inform. 2021 Jan 27;9(1):e24207
pubmed: 33400679
Nat Commun. 2020 Oct 6;11(1):5033
pubmed: 33024092
Med J Islam Repub Iran. 2020 Jun 25;34:68
pubmed: 32974234
Int J Epidemiol. 2020 Jun 1;49(3):717-726
pubmed: 32086938
Int J Antimicrob Agents. 2020 Mar;55(3):105924
pubmed: 32081636
Acad Emerg Med. 2020 Jun;27(6):461-468
pubmed: 32311790
J Biomed Phys Eng. 2021 Oct 01;11(5):653-662
pubmed: 34722410
J Thorac Dis. 2019 Mar;11(Suppl 4):S574-S584
pubmed: 31032076
Lancet Infect Dis. 2020 Apr;20(4):425-434
pubmed: 32105637
Sci Rep. 2020 Oct 30;10(1):18716
pubmed: 33127965
BMJ. 2020 Apr 7;369:m1328
pubmed: 32265220
Diabetes Res Clin Pract. 2020 Apr;162:108142
pubmed: 32278764
J Med Internet Res. 2020 Nov 11;22(11):e23128
pubmed: 33035175
Scand J Trauma Resusc Emerg Med. 2020 Oct 27;28(1):106
pubmed: 33109234
BMC Med Res Methodol. 2020 Oct 20;20(1):262
pubmed: 33081694
Crit Rev Oncol Hematol. 2020 Jun;150:102972
pubmed: 32344317
Int J Environ Res Public Health. 2020 Apr 16;17(8):
pubmed: 32316191
Eur Respir J. 2020 Aug 20;56(2):
pubmed: 32616597
Front Public Health. 2021 May 12;9:626697
pubmed: 34055710
Eur Heart J. 2017 Feb 14;38(7):500-507
pubmed: 27252451
J Med Syst. 2020 Aug 1;44(9):156
pubmed: 32740678
Procedia Comput Sci. 2021;179:871-877
pubmed: 33643495
BMC Med Inform Decis Mak. 2016 Mar 29;16:39
pubmed: 27025458
Ann Med Surg (Lond). 2020 Nov;59:207-216
pubmed: 33042536
Cardiovasc Revasc Med. 2021 Jan;22:22-28
pubmed: 32591310