tetano
Editor, Senior Moderator
J Med Internet Res
. 2021 Mar 18.
doi: 10.2196/25852. Online ahead of print.
Prediction models for clinical severity of COVID-19 patients using multi-center clinical data in Korea
Bumjo Oh[SUP] 1 [/SUP], Suhyun Hwangbo[SUP] 2 [/SUP], Taeyeong Jung[SUP] 2 [/SUP], Kyungha Min[SUP] 1 [/SUP], Chanhee Lee[SUP] 2 [/SUP], Catherine Apio[SUP] 2 [/SUP], Hyejin Lee[SUP] 3 [/SUP], Seungyeoun Lee[SUP] 4 [/SUP], Min Kyong Moon[SUP] 5 [/SUP], Shin-Woo Kim[SUP] 6 [/SUP], Taesung Park[SUP] 7 [/SUP]
Affiliations
Abstract
Background: There is limited information describing present characteristics and dynamic clinical changes that occur in patients with severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2) infection during the early phase of illness.
Objective: The objective is to develop and validate machine-learning model based on clinical features for severity risk assessment and triage for COVID-19 patients at hospital admission.
Methods: This is a retrospective cohort of multicenter COVID-19 patients released from quarantine until April 30th, 2020 in Korea. A total of 5,628 patients were used to train and validate the models that predict the clinical severity and duration of hospitalization, where clinical severity score was defined in 4 levels: mild, moderate, severe, and critical.
Results: The proportion of patients in the mild, moderate, severe, and critical levels were 79.5% (4455/5601), 5.9% (330/5601), 9.1% (512/5601), and 5.4% (301/5601), respectively. As risk factors for predicting critical patients, older age, shortness of breath, higher white blood cell, lower hemoglobin, lower lymphocyte, and lower platelet count were selected. Three prediction models were built to classify clinical severity levels. For example, the prediction model with 6 variables showed the predictive power of 0.93 or higher for the area under the receiver operating characteristic curve (AUC). Based on these models, a web-based nomogram was developed (http://statgen.snu.ac.kr/covid19/nomogram/maxcss/).
Conclusions: Our prediction models, along with the web-based nomogram are expected to be useful to access the onset of severe and critical illness among COVID-19 patients and triage at hospital admission.
. 2021 Mar 18.
doi: 10.2196/25852. Online ahead of print.
Prediction models for clinical severity of COVID-19 patients using multi-center clinical data in Korea
Bumjo Oh[SUP] 1 [/SUP], Suhyun Hwangbo[SUP] 2 [/SUP], Taeyeong Jung[SUP] 2 [/SUP], Kyungha Min[SUP] 1 [/SUP], Chanhee Lee[SUP] 2 [/SUP], Catherine Apio[SUP] 2 [/SUP], Hyejin Lee[SUP] 3 [/SUP], Seungyeoun Lee[SUP] 4 [/SUP], Min Kyong Moon[SUP] 5 [/SUP], Shin-Woo Kim[SUP] 6 [/SUP], Taesung Park[SUP] 7 [/SUP]
Affiliations
- PMID: 33822738
- DOI: 10.2196/25852
Abstract
Background: There is limited information describing present characteristics and dynamic clinical changes that occur in patients with severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2) infection during the early phase of illness.
Objective: The objective is to develop and validate machine-learning model based on clinical features for severity risk assessment and triage for COVID-19 patients at hospital admission.
Methods: This is a retrospective cohort of multicenter COVID-19 patients released from quarantine until April 30th, 2020 in Korea. A total of 5,628 patients were used to train and validate the models that predict the clinical severity and duration of hospitalization, where clinical severity score was defined in 4 levels: mild, moderate, severe, and critical.
Results: The proportion of patients in the mild, moderate, severe, and critical levels were 79.5% (4455/5601), 5.9% (330/5601), 9.1% (512/5601), and 5.4% (301/5601), respectively. As risk factors for predicting critical patients, older age, shortness of breath, higher white blood cell, lower hemoglobin, lower lymphocyte, and lower platelet count were selected. Three prediction models were built to classify clinical severity levels. For example, the prediction model with 6 variables showed the predictive power of 0.93 or higher for the area under the receiver operating characteristic curve (AUC). Based on these models, a web-based nomogram was developed (http://statgen.snu.ac.kr/covid19/nomogram/maxcss/).
Conclusions: Our prediction models, along with the web-based nomogram are expected to be useful to access the onset of severe and critical illness among COVID-19 patients and triage at hospital admission.