Machine Learning Prediction Models for COVID-19 ICU Mortality: Model Development and Validation

Machine Learning Prediction Models for COVID-19 ICU Mortality: Model Development and Validation | Research Square window.SnipcartSettings = { analytics: { enabled: false } }; (function() { var accessVector = localStorage.getItem('access_vector') || ''; window.dataLayer = window.dataLayer || []; if (accessVector) { window.dataLayer.push({ user: { profile: { profileInfo: { snid: accessVector } } } }); } })(); (function(w,d,s,l,i){w[l]=w[l]||[];w[l].push({'gtm.start':new Date().getTime(),event:'gtm.js'});var f=d.getElementsByTagName(s)[0],j=d.createElement(s),dl=l!='dataLayer'?'&l='+l:'';j.async=true;j.src='https://www.googletagmanager.com/gtm.js?id='+i+dl;f.parentNode.insertBefore(j,f);})(window,document,'script','dataLayer','GTM-K279D39R'); Browse Preprints In Review Journals COVID-19 Preprints AJE Video Bytes Research Tools Research Promotion AJE Professional Editing AJE Rubriq About Preprint Platform In Review Editorial Policies Our Team Advisory Board Help Center Sign In Submit a Preprint Cite Share Download PDF Research Article Machine Learning Prediction Models for COVID-19 ICU Mortality: Model Development and Validation Ali Heidarnejad Maleki, Zeinab Khanali, Ali Khayami, Samad Ghodarti This is a preprint; it has not been peer reviewed by a journal. https://doi.org/ 10.21203/rs.3.rs-8173098/v1 This work is licensed under a CC BY 4.0 License Status: Posted Version 1 posted You are reading this latest preprint version Abstract Introduction: Predicting the severity and outcome of COVID-19 is a challenging task. This study investigated the potential of predicting mortality using the SOFA score, CT scan findings based on the CO-RADS system, and biomarkers (including IL-6 and LDH) in intensive care unit (ICU)-admitted patients with COVID-19. Additionally, we developed multivariable models to enhance prognostic accuracy. Materials and methods This retrospective cohort study was conducted on 426 COVID-19 patients admitted to the ICU of a tertiary hospital in Zanjan, Iran, from March to November 2020. The data were collected from patients' medical records. The correlation between variables and mortality was analyzed, and the predictability of mortality was assessed using the receiver operating characteristic (ROC) curve. Cut-off points, sensitivity, and specificity were determined. Conventional logistic regression methods and four machine learning (ML) algorithms were employed to develop mortality prediction models for ICU patients with COVID-19 using Python. The performance of these machine learning models was measured by the area under the receiver operating characteristic curve (AUC). The internal validation of these ML-based models was performed using an integrated 10-fold stratified cross-validation with bootstrap. Results The mortality rate was 47.1% (n = 200). The mean SOFA (5.23 vs. 3.58, p < 0.001) and CO-RADS (5.54 vs. 5.03, p 18.95 pg/mL, 65.5% sensitivity, 93.4% specificity) and LDH (AUC: 0.737, cut-off: >437.5 U/L, 63.0% sensitivity, 68.6% specificity) yielded the highest predictability for mortality, followed by SOFA (AUC: 0.701, cut-off: >3, 80.0% sensitivity, 46.5% specificity) and among comorbidities, hypertension and diabetes exhibited significant correlation with mortality risk (p = 0.001 and p = 0.038, respectively). Using logistic regression methods and four machine learning (ML) algorithms, a six-factor model was developed involving age, IL-6, LDH, SOFA, CO-RADS, and ESR. This model yielded the best AUC in the adaptive boosting (AdaBoost) algorithm (AUC: 0.960, sensitivity: 0.915, specificity: 0.885). According to the internal validation and calibration plots (Fig. 3), the closest agreement between predicted and observed probabilities was shown in the categorical boosting (CatBoost) algorithm (Brier score: 0.086, AUC: 0.952). Conclusions IL-6 and LDH showed the highest predictive value for mortality among biomarkers, and the SOFA score showed predictability compared to the CT-based CO-RADS system. The six-factor AdaBoost model showed the best performance in predicting the mortality risk. COVID-19 ICU SOFA score IL-6 LDH CO-RADS mortality prediction models machine learning Figures Figure 1 Figure 2 Figure 3 Introduction In December 2019, a new coronavirus was identified as the cause of pneumonia in Wuhan, China. In February 2020, the World Health Organization (WHO) named the coronavirus disease 2019 (COVID-19) ( 1 ). According to the World Health Organization, as of October 13, 2024, more than 776 million verified cases and more than 7 million deaths have been reported worldwide ( 2 ). Even with coronavirus vaccinations, COVID-19 waves can still impact lives due to mutations that have occurred since the disease emerged ( 3 ). Studies indicate that 15–30% of hospitalized COVID-19 patients require admission to an intensive care unit (ICU), and the mortality rate of ICU-admitted patients ranges from 30% to over 50%, depending on the healthcare system and patient demographics ( 4 , 5 ). Although the mortality rate of COVID-19 is as low as 3%, its rapid transmission resulted in a considerable number of deaths ( 6 ). Reverse transcriptase polymerase chain reaction (RT-PCR) is the gold standard diagnostic test for COVID-19. Following these studies, CT-scan signs of pulmonary involvement were incorporated into the diagnostic protocol for this disease in China. In addition, ARDS can be a complication of COVID-19, leading to disease progression and increased mortality ( 7 , 8 ). The risk stratification was often based on lung involvement, making CT scans valuable diagnostic tools during the pandemic ( 9 , 10 ). A Lung CT scan helps diagnose COVID-19 and can be crucial in evaluating the response to treatment and disease progression ( 11 ). The COVID-19 Reporting and Data System (CO-RADS), introduced by the Dutch Radiological Society, categorizes pulmonary involvement in COVID-19 based on non-enhanced chest CT scans. It effectively predicted the virus in patients with moderate to severe symptoms ( 12 ). The significant advantage of this classification is the ease of use. There was a need for disease severity scoring systems, such as SOFA and APACHE, to improve patient outcomes and optimize resource use. The Sequential Organ Failure Assessment (SOFA) is a crucial tool for evaluating multi-organ failure and is recognized as a predictive factor for outcomes in various diseases ( 13 , 14 ). SOFA scoring is based on assessing the function of six organs: blood circulation, respiratory system, liver, kidney, central nervous system (CNS), and coagulation system. The score assigned to each organ can range between 3 and 8. SOFA is a quick and accessible tool for rapidly evaluating hospitalized patients in acute conditions ( 13 ). In the present study, biomarkers were also investigated, including serum levels of interleukin 6 (IL-6), lactate dehydrogenase (LDH), sedimentation rate (ESR), C-reactive protein (CRP), white blood cell (WBC) count, lymphocyte count, platelet count, total bilirubin, and creatinine. As initial evaluation and treatment response surveillance in COVID-19 patients can prevent mortality and complications, identifying predictors of disease severity is crucial in these patients. Considering the ease of use of SOFA and the availability of CT scans as a diagnostic tool for COVID-19, these two methods are hypothetically helpful in evaluating the risk of mortality and COVID-19 severity. Recent studies have investigated the correlation between chest CT findings, SOFA scores, and mortality in COVID-19 patients: A chest CT scan alone is insufficient to predict mortality in COVID-19 ARDS patients, and it is advisable to utilize the SOFA score for multi-organ assessment ( 15 ). The SOFA scoring system, along with imaging and other clinical findings, was recommended for predicting mortality in patients with COVID-19 pneumonia ( 16 ). This study aimed to investigate the predictive properties of various clinical parameters. While existing research has explored the prognostic value of clinical parameters, there is a limited understanding of integrating the predictive value of clinical, laboratory, and imaging findings to enhance COVID-19 mortality prediction in the intensive care setting. We designed ML-based models to develop mortality prediction performance and conducted internal validation for each model. The designed prediction models incorporate biomarkers and clinical findings, and depending on the availability of clinical data, these models may serve as valuable prognostic tools for the early identification of critically ill patients. This plays a crucial role in optimizing the allocation of medical resources, guiding patient triage, personalized treatment decision-making, and monitoring the progression of COVID-19. Methods The current study was a retrospective cohort study conducted among patients with COVID-19 who were admitted to the intensive care unit (ICU) department of a tertiary hospital in Zanjan, Iran, from March to November 2020. The study population was selected using census sampling. The inclusion criteria were an age of 18 years or older and a documented diagnosis of COVID-19, as confirmed by RT-PCR or CT scan. The exclusion criteria were pregnancy and incomplete patient files. Data were collected from patients' records and imported into a checklist. Data included demographic data (age and gender), past medical history (hypertention, diabetes, ischemic heart diseases (IHD), chronic obstructive pulmonary diseases (COPD), cancer, congestive heart failure (CHF), cerebrovascular accidents (CVA), hyperlipidemia (HLP), chronic kidney diseases (CKD)), vital signs (heart rate, respiratory rate, temperature, systolic and diastolic blood pressure, mean arterial pressure), laboratory findings (serum LDH, CRP, ESR, IL-6, WBC count, lymphocytes count, serum creatinine, total bilirubin, platelet count (Plt), and arterial oxygen pressure [PaO 2 ]), spirometry findings (fraction of inspired oxygen [FiO 2 ]), Oxygenation index (OI: PaO 2 /FiO 2 ), consciousness level based on the Glasgow Coma Scale (GCS). The SOFA score was calculated from the patient records. The CO-RADS was evaluated based on the archived CT scans of the included patients. Study instruments The SOFA score was calculated based on the evaluation of the function of six organs ( 17 ). In each organ evaluation, normal function is assigned a score of zero, and the SOFA score increases with an increased level of dysfunction in the organ. The criteria for evaluating the respiratory system are PaO 2 /FiO 2 and the need for respiratory support. Based on these criteria, a PaO 2 /FiO 2 above 400 is scored as zero, a PaO 2 /FiO 2 between 301 and 400 is scored as one, a PaO 2 /FiO 2 between 201 and 300 is scored as two, a PaO 2 /FiO 2 between 101 and 200 with respiratory support is scored as three, and a PaO 2 /FiO 2 below 100 with respiratory backing is scored as four. Platelet count above 150 (1000/µl) is scored zero, while platelet counts of 101–150 (1000/µl), 51–100 (1000/µl), 21–50 (1000/µl), and equal or lower than 20 (1000/µl) are scored, one, two, three, and four, respectively. Serum bilirubin below 1.2 mg/dL is scored zero, while serum bilirubin levels 1.2–1.9 mg/dL, 2.0-5.9 mg/dL, 6.0-11.9 mg/dL, and above 12 mg/dL are scored one, two, three, and four, respectively. A mean arterial pressure (MAP) greater than or equal to 70 mmHg is scored zero. In comparison, MAP below 70 mmHg, need for 5 mg dopamine or any dose of dobutamine, need for more than 5 mg dopamine or epinephrine or norepinephrine less than or equal to 0.1, need for more than 15 mg dopamine or epinephrine or norepinephrine less than or equal to 0.1 are scored, one, two, three, and four, respectively. Serum creatinine levels below 1.2 mg/dL are scored as zero. In contrast, serum creatinine levels of 1.2–1.9 mg/dL, 2.0-3.4 mg/dL, 3.5–4.9 mg/dL, or urine output less than 500 mL/d, serum creatinine levels more than 5.0 mg/dL, or urine output less than 200 mL are scored as one, two, three, and four, respectively. A GCS of 15 is scored zero, while GCS scores of 13–14, 10–12, 6–9, and less than six are scored one, two, three, and four, respectively ( 17 ). CT interpretations and the CO-RADS scoring were performed using the CO-RADS criteria by a qualified board-certified radiologist. The CO-RADS score was zero if the CT scan quality prevented the radiologist from categorizing any scores ( 18 ). A score of one is given for a routine CT scan or findings attributed to non-infectious disorders. Score two is given to a CT scan with conclusions related to infectious diseases but not COVID-19 (bronchitis, bronchiolitis, bronchopneumonia, centrilobular GGO, or pulmonary abscess). Score three is given to CT scan with unclear findings for COVID-19 (perihilar GGO, homogenous and widespread GGO, GGO related to interlobular thickening, and pneumonia patterns in case of the absence of other COVID-19 findings. A score of four is given to a CT scan with suspicious COVID-19 findings that may overlap with those of other viral pneumonias. A score of 5 is given to a CT scan with highly suggestive findings of COVID-19 ( 18 ). Statistical analysis Initial statistical analyses were performed using IBM SPSS Statistics version 26 (IBM Corp., Armonk, NY, USA). The results of these analyses are presented in Tables 1–3 and Fig. 1. Advanced modeling and machine learning analyses were conducted using Python version 3.10 (Python Software Foundation, Wilmington, DE, USA), incorporating packages such as scikit-learn (sklearn), seaborn, matplotlib, pandas, and NumPy. The results of the Python-based analyses are presented in Tables 4 and 5, as well as Figs. 2 and 3. Table 1 Demographic and clinical characteristics Characteristic Overall (n = 426) Discharged (N = 226) Deceased (n = 200) p-value Age (y), mean (SD) years 1 61.51 (16.90) 57.57 (18.28) 65.96 (13.96) < 0.001* Gender, male, no (%) 2 224 (52.60%) 121 (53.53%) 103 (51.50%) 0.714 Comorbidities, no (%) Hypertension 2 174 (40.8%) 76 (33.62%) 98 (49%) 0.001* Diabetes mellitus 2 96 (22.53%) 42 (18.58%) 54 (27%) 0.038* Ischemic heart disease 2 43 (10.09%) 20 (8.85%) 23 (11.5%) 0.365 Chronic Obstructive Pulmonary Disease 2 31 (7.28%) 15 (6.64%) 16 (8%) 0.589 Neoplasia 2 16 (3.76%) 5 (2.21%) 11 (5.5%) 0.075 Congestive Heart Failure 2 10 (2.35%) 7 (3.1%) 3 (1.5%) 0.277 Cerebrovascular accident 2 13 (3.05%) 8 (3.54%) 5 (2.5%) 0.533 Hyperlipidemia 2 17 (3.99%) 8 (3.54%) 9 (4.5%) 0.613 Chronic kidney disease 2 13 (3.05%) 5 (2.21%) 8 (4%) 0.284 Mechanical ventilation, positive, no (%) 2 48 (11.27%) 15 (6.64%) 33 (16.5%) 0.001* SARS-CoV-2 RT-PCR result, positive, no (%) 2 292 (68.54%) 132 (58.40%) 160 (80%) < 0.001* CO-RADS category, no (%) 1 5.27 (1.12) 5.03 (1.19) 5.54 (0.96) < 0.001* 0 0 (0%) 0 (0%) 0 (0%) 1 0 (0%) 0 (0%) 0 (0%) 2 0 (0%) 0 (0%) 0 (0%) 3 46 (10.79%) 32 (14.15%) 14 (7.00%) 4 86 (20.18%) 61 (26.99%) 25 (12.50%) 5 2 (0.46%) 1 (0.44%) 1 (0.50%) 6 292 (68.54%) 132 (58.40%) 160 (80.00%) SD Standard deviation, no number, SOFA sequential organ failure assessment, CO-RADS COVID-19 Reporting and Data System, *significant differences (p‑value < 0.05), 1/ Mann-Whitney U test, 2/ Chi‑square test or exact Fisher test Table 3 Variables as potential predictors of mortality *PPV: Positive predictive value, **NPV: Negative predictive value ROC Risk factor cutoff characterization Cutoff univariate logistic regression Variables AUC 95% CI P value Cut-off Sensitivity Specificity PPV NPV Value OR 95% CI P value Age 0.633 0.581–0.686 < 0.001 50 0.87 0.347 0.54 0.74 ≤ 50 ref 50 3.366 2.075–5.459 CRP 0.597 0.535–0.658 0.002 70.0 (mg/L) 0.389 0.814 0.57 0.68 ≤ 70 (mg/L) ref 70 (mg/L) 2.788 1.736–4.478 IL-6 0.761 0.711–0.810 < 0.001 18.95 (pg/mL) 0.655 0.934 0.90 0.75 ≤ 18.95 (pg/mL) ref 18.95 (pg/mL) 26.706 14.665–48.634 LDH 0.737 0.690–0.783 < 0.001 437.5 (U/L) 0.630 0.686 0.64 0.68 ≤ 437.5 (U/L) ref 437.5 (U/L) 3.717 2.488–5.554 WBC 0.555 0.500–0.610 0.049 10.15 (×10 9 /L) 0.540 0.571 0.52 0.58 ≤ 10.15 (×10 9 /L) ref 0.022 > 10.15 (×10 9 /L) 1.561 1.064–2.290 Bilirubin (Total) 0.582 0.528–0.636 0.003 1.15 (mg/dL) 0.465 0.699 0.58 0.60 ≤ 1.15 (mg/dL) ref 1.15 (mg/dL) 2.020 1.357–3.005 PaO 2 0.612 0.559–0.666 68.45 (mmHg) ref < 0.001 ≤ 68.45 (mmHg) 2.416 1.634–3.574 PaO 2 /FiO 2 (OI) 0.643 0.591–0.695 325.9524 ref < 0.001 ≤ 325.9524 2.416 1.634–3.574 SOFA 0.701 0.652–0.751 < 0.001 3 0.800 0.465 0.64 0.70 ≤ 3 ref 3 4.221 2.759–6.459 CO-RADS 0.608 0.555–0.661 < 0.001 4 0.807 0.410 0.57 0.70 ≤ 4 ref 4 3.399 2.125–5.439 *PPV: Positive predictive value, **NPV: Negative predictive value Table 4 Area under the curve of multivariable prediction models Models AUC LR GBDT AdaBoost XGBoost CatBoost IL-6 + LDH 0.801 0.896 0.874 0.887 0.892 Age + IL-6 + LDH 0.810 0.901 0.891 0.877 0.895 Age + IL-6 + SOFA 0.804 0.865 0.868 0.847 0.869 IL-6 + LDH + SOFA 0.830 0.907 0.897 0.887 0.902 Age + IL-6 + SOFA + CO-RADS 0.855 0.884 0.884 0.872 0.892 Age + IL-6 + LDH + SOFA 0.831 0.910 0.906 0.884 0.908 Age + IL-6 + LDH + SOFA + CO-RADS 0.873 0.928 0.919 0.914 0.925 Age + IL-6 + LDH + SOFA + CO-RADS + ESR 0.877 0.953 0.960 0.941 0.952 LR: Logistic regression, GBDT: gradient boosting decision tree, AdaBoost: adaptive boosting, XGBoost: eXtreme Gradient Boosting, CatBoost: categorical boosting, AUC: area under the receiver operating characteristic curve Table 5 Parameters in the various model for (Age + IL-6 + LDH + SOFA + CO-RADS + ESR) prediction model LR GBDT AdaBoost XGBoost CatBoost AUC 0.877 0.953 0.960 0.941 0.952 95% CI of AUC 0.840–0.908 0.933–0.970 0.941–0.974 0.919–0.961 0.932–0.969 Brier score 0.141 0.084 0.163 0.102 0.086 Youden index 0.594 0.596 0.480 0.624 0.523 Sensitivity 0.690 0.870 0.915 0.875 0.875 Specificity 0.898 0.916 0.885 0.889 0.903 PPV 0.857 0.902 0.876 0.875 0.888 NPV 0.766 0.888 0.922 0.889 0.891 LR: Logistic regression, GBDT: gradient boosting decision tree, AdaBoost: adaptive boosting, XGBoost: eXtreme Gradient Boosting, CatBoost: categorical boosting, AUC: area under the receiver operating characteristic curve, CI: confidence interval, PPV: Positive predictive value, NPV: Negative predictive value The Kolmogorov-Smirnov test was used to assess the normality of quantitative variables. Descriptive statistics were presented using frequencies and percentages for qualitative variables, and means, standard deviations, medians, and interquartile ranges for quantitative variables. The independent t-test was used to compare the means of normally distributed quantitative variables between study groups. In contrast, the Mann-Whitney test was used to compare the median of non-normally distributed variables between groups. The chi-square or Fisher's exact test was used to compare qualitative variables between groups. The Pearson's correlation coefficient test was used to assess the correlation of quantitative variables in this study. The logistic regression test investigated the relationship between the studied variables and the disease outcome (mortality rate). The results of this test were expressed as a p-value, odds ratio (OR), and 95% confidence interval for OR. The Receiver Operating Characteristic (ROC) test was used to determine the cut-off point for determining mortality. The results of this statistical analysis were expressed using the area under the curve (AUC) and the 95% confidence interval for the AUC. Youden's index was also used to determine the cut-off point, sensitivity, and specificity. The Combined ROC curve was analyzed to integrate the predictability of multiple variables and introduced nine multi-variable models. We aimed to introduce the highest AUC and predictability, while involving the minimum number of variables. This has now been clarified in the revised manuscript. The multicollinearity among the predictors in the multivariable ROC model was assessed using the Variance Inflation Factor (VIF) in Python. All VIF values were below 1.1, indicating the absence of significant collinearity between the variables. A significance level of less than 0.05 was considered in this study. Development and validation of the Models We employed a conventional logistic regression method and popular machine learning classification algorithms, including adaptive boosting (AdaBoost), gradient boosting decision tree (GBDT), eXtreme Gradient Boosting (XGBoost), and categorical boosting (CatBoost) to develop predictive models and perform the internal validation via combining the 10-fold cross-validation and bootstrap resampling. The models were designed to involve two to six parameters, aiming to create the best AUC with the minimum number of parameters. The best model was determined according to the AUC comparison. Internal validation was performed using the combination of 10-fold cross-validation and a bootstrap resampling approach to ensure robust estimation of model performance and mitigate potential overfitting. For each fold, models were trained on 90% of the random data and evaluated on the remaining 10%. Within each fold, we applied bootstrap resampling with 1,000 iterations to the test set to estimate confidence intervals for performance metrics, including AUC, sensitivity, and specificity. Predicted probabilities across all folds were then pooled to construct an overall ROC. Interpretation and evaluation of the risk prediction model were performed using calibration curves. The Youden index, sensitivity, and specificity at the cut-off point were determined. These analyses were performed using Python (version 3.9) with scikit-learn, XGBoost, and CatBoost packages. Results Study participants A total of 426 COVID-19 patients admitted to the ICU were studied. Of these, 226 (52.9%) patients were discharged, and 200 (47.1%) patients were deceased. The demographic and clinical features of the discharged and deceased groups are recorded in Table 1 . The mean age was 61.51 years, which was significantly higher in the deceased group compared to the discharged group (65.96 vs. 57.57; P < 0.001). The study population consisted of 224 (52.6%) male and 202 (47.4%) female patients. Male patients were more affected; however, there was no significant difference in gender between the deceased and discharged patients. RT-PCR testing was reported as positive for COVID-19 in 292 (68.54%) patients. Hypertension was the most common comorbidity (40.8%), followed by diabetes mellitus (22.53%), in the overall and the two groups. Among the 48 patients who required mechanical ventilation support, 33 were deceased, and 15 were discharged. There were significant differences between the discharged and deceased groups in terms of the prevalence of hypertension (p = 0.001), diabetes mellitus comorbidities (p = 0.001), and the number of patients who needed mechanical ventilation (p = 0.001). The mean SOFA score on admission was 4.35, indicating a significant difference between the discharged and deceased groups (3.58 vs. 5.23; P < 0.001). The CO-RADS score also showed a significant difference between the discharged and deceased groups (5.03 vs. 5.54; P < 0.001). In Table 2 , the mean and standard deviation values are recorded for the measured parameters. The odds ratio of parameters in association with mortality was calculated using logistic regression analysis, and the OR of the following parameters was higher than 1 (OR > 1) and significant (P < 0.05): 1. HR (OR: 1.017, P value < 0.001), 2. RR (OR:1.053, P value = 0.005), 3. IL-6 (OR: 1.023, P value < 0.001), 4. ESR (OR:1.011, P value < 0.001), 5. CRP (OR:1.011, P value < 0.001), 6. LDH (OR: 1.010, P value < 0.001), 7. Total bilirubin (OR: 1.319, P value: 0.020), 8. Creatinine (OR: 1.147, P value: 0.167), 9. SOFA score (OR: 1.345, P value < 0.001), 10. CO-RADS (OR: 1.534, P value < 0.001). Table 2 Clinical and laboratory measurements and their association with mortality Discharged Deceased Univariate logistic regression Variables Mean (SD*) Mean (SD) P value OR 95% CI P value Clinical indices Hear rate 1 85.71 (17.369) 91.45 (20.031) 0.002* 1.017 1.006–1.027 < 0.001* Temperature 2 (°C) 36.808 (0.540) 36.762 (0.536) 0.259 0.851 0.595–1.217 < 0.001* Respiratory Rate** 2 (210/167) 21.13 (5.421) 22.86 (6.208) 0.002* 1.053 1.016–1.091 0.005* SBP 1 (mmHg) 127.59 (21.888) 127.59 (25.512) 0.997 1.000 0.992–1.008 0.997 DBP 1 (mmHg) 78.06 (14.452) 77.50 (17.296) 0.716 0.998 0.086–1.010 0.712 MAP 1 (mmHg) 94.572 (15.666) 94.191 (18.886) 0.822 0.999 0.088–1.010 0.820 GCS 2 14.47 (1.424) 13.28 (2.841) < 0.001* 0.768 0.692–0.853 < 0.001* Oxygen indices PaO 2 2 (mmHg) 73.732 (22.891) 66.297 (15.605) < 0.001* 0.978 0.966–0.990 < 0.001* FiO 2 2 (%) 26.59 (20.307) 34.03 (29.397) 0.003* 1.012 1.004–1.020 0.003* PaO 2 /FiO 2 (OI) 2 321.147 (94.186) 269.308 (103.142) < 0.001* 0.995 0.992–0.997 < 0.001* Laboratory indices WBC (× 10 3 cells/µL) 2 11.257 (17.484) 11.739 (8.432) 0.049* 1.003 0.989–1.017 0.725 Lymphocyte cells (× 10 3 cells/µL) 2 1.556 (9.283) 1.529 (5.078) 0.434 0.999 0.975–1.025 0.961 Lymphocyte (%) 2 9.413 (7.157) 10.191 (9.532) 0.641 1.011 0.988–1.035 0.340 Platelet cells (× 10 3 cells/µL) 2 225.94 (101.130) 210.82 (95.561) 0.087 0.998 0.996-1.000 0.116 IL-6 2 (pg/mL) 15.656 (67.214) 52.822 (58.862) < 0.001* 1.023 1.016–1.031 < 0.001* ESR 2 (mm/hr) 56.527 (21.289) 70.979 (48.008) 0.118 1.011 1.006–1.017 < 0.001* CRP 2 *** (mg/L) 48.694 (33.061) 70.659 (61.235) 0.002* 1.011 1.006–1.016 < 0.001* LDH 2 (U/L) 393.85 (83.974) 482.85 (103.064) < 0.001* 1.010 1.007–1.012 < 0.001* Bilirubin (Total) 2 (mg/dL) 1.134 (0.879) 1.356 (0.995) 0.003* 1.319 1.044–1.666 0.020* Creatinine 2 (mg/dL) 1.208 (0.938) 1.353 (1.155) 0.090 1.147 0.944–1.393 0.167 Scoring systems SOFA 2 3.58 (2.034) 5.23 (2.766) < 0.001* 1.345 1.226–1.477 < 0.001* CO-RADS 2 5.03 (1.194) 5.54 (0.961) < 0.001* 1.534 1.275–1.844 < 0.001* *Significant differences (p‑value < 0.05), **Respiratory rate data were available for 210 discharged and 167 deceased patients, ***CRP data were available for 226 discharged and 144 deceased patients. Other parameters were recorded for all 226 discharged and 200 deceased patients. SD Standard Deviation, OR Odds Ratio, CI Confidence interval, SBP systolic blood pressure, DBP diastolic blood pressure, MAP mean arterial pressure, GCS Glasgow Coma Scale, PaO 2 arterial partial pressure of oxygen, FiO 2 fraction of inspiratory oxygen concentration, OI oxygenation index, WBC white blood cell count, IL-6 interleukin 6, ESR erythrocytes sedimentation rate, CRP c-reactive protein, LDH lactate dehydrogenase, SOFA sequential organ failure assessment, CO-RADS COVID-19 Reporting and Data System, 1/ T-test, 2/ Mann–Whitney U test. Findings indicate that a one-point increase in SOFA and CO-RADS scores corresponded to a 34.5% and 53.4% increase in mortality risk, respectively. A one-point rise in GCS was associated with a 23.2% decrease in mortality risk. In addition to the clinical findings, the odds ratios of inflammatory markers are listed in Table 2 . According to Table 3 and ROC curves in Fig. 1 , the cut-off points for mortality were measured for the following parameters using the Youden index: Age (Fig. 1 A) (> 50 years, AUC:0.633, 87% sensitivity, 34.7% specificity), CRP (Fig. 1 B) (> 70 mg/L, AUC:0.597, 38.9% sensitivity, 81.4% specificity), IL-6 (Fig. 1 C) (> 18.95 pg/mL, AUC: 0.761, 65.5% sensitivity, 93.4% specificity), LDH (Fig. 1 D) (> 437.5 U/L, AUC: 0.737, 63% sensitivity, 68.6% specificity), WBC (Fig. 1 E) (> 10.15×10 9 /L, AUC: 0.555, 54% sensitivity, 57.1% specificity), Bilirubin (Total) (Fig. 1 F) (> 1.15 mg/dL, AUC: 0.582, 46.5% sensitivity, 69.9% specificity), PaO 2 (Fig. 1 G) (≤ 68.45 mmHg, AUC: 0.612, 64.5% sensitivity, 57.1% specificity), PaO 2 /FiO 2 (Oxygenation index: OI) (Fig. 1 H) (≤ 325.95, AUC: 0.643, 50.9% sensitivity, 73.5% specificity), SOFA (Fig. 1 I) (> 3, AUC: 0.701, 80% sensitivity, 46.5% specificity) (Fig. 1 J) (> 18.5, AUC: 0.662, 66% sensitivity, 55.8% specificity), CO-RADS (Fig. 1 K) (> 4, AUC:0.608, 80.7% sensitivity, 41% specificity). Machine learning models performance and validation The AUC of multi-factor models was measured using logistic regression and four ML-based algorithms. Models with an AUC > 0.800 were recorded in Table 4 , and related ROCs are demonstrated in Fig. 2 . According to Table 4 , the six-factor model, which included age, IL-6, LDH, SOFA, CO-RADS, and ESR, presented the highest AUC in each setting compared to the other models. In terms of prediction performance, the results of the logistic regression and four machine learning algorithms on the test set indicated AUCs of 0.960 for AdaBoost, 0.953 for GBDT, 0.952 for CatBoost, 0.941 for XGBoost, and 0.877 for logistic regression. The AUC, 95% CI, Brier score, Youden index, sensitivity, specificity, PPV, and NPV values of each model in the test data set are listed in Table 5 . The calibration curves for internal validation for each model are shown in Fig. 3 . The AdaBoost model demonstrated the best performance with an AUC of 0.960 (95% CI: 0.941–0.974). The Youden index was 0.480, and sensitivity and specificity at the optimum cut-off point were reported as 0.915 and 0.885, respectively. According to combined internal validation, the GBDT algorithm demonstrated the best calibration, as indicated by the Brier score (0.084). Discussion We included 426 ICU-admitted COVID-19 patients and investigated their mortality, clinical features, and biomarkers in two groups: the deceased and discharged groups. The male-to-female ratio was 1.1:1 (224 males vs. 202 females), and the mortality rate was 47.1% (n = 200). Hypertension and diabetes mellitus were the most common comorbidities in both groups, and there was a significant correlation between these comorbidities and the mortality rate. In a systematic review and meta-analysis of the findings of 10 articles (76,993 patients), the prevalence of hypertension, cardiovascular diseases, cancer, chronic kidney failure, smoking, and diabetes was significantly higher in COVID-19 patients compared to non-affected patients ( 19 ). In a systematic review of the findings of 114 studies (310,494 patients), among the 72 predictive variables, advanced age, hypertension, and diabetes were reported as the most important predictors of mortality (based on the 10% cut-off point). Diabetes was identified as an independent risk factor for mortality in COVID-19 patients ( 20 ). In a community study of 3,862,012 COVID-19 patients in England, cardiovascular diseases, diabetes, a history of steroid treatment, obesity, chronic kidney disease, chronic obstructive pulmonary disease, and immunodeficiency diseases were significantly associated with mortality from COVID-19 in patients aged 70 years or older ( 21 ). In our study, deceased patients were significantly older than the discharged group. The mortality risk for patients older than 50 was 3.366 times higher than that for younger patients. In agreement with our findings, a systematic review study encompassing 49 studies, which included more than 587,000 COVID-19 patients, reported that advanced age (defined as being older than 50 years) and comorbidities significantly increased the risk of in-hospital mortality in COVID-19 patients. Of note, the risk of hospital mortality in patients over 60 years old was 23 times higher than that of patients who were less than 60 years old ( 22 ). In another systematic review and meta-analysis of the data from five countries, China, Italy, Spain, England, and the USA (New York state) (611,583 patients in total), a significant association was reported between age and mortality rate and nosocomial mortality in patients with COVID-19 ( 23 ). The mortality rate in ICU-admitted COVID-19 patients is variable depending on the region and health system. It has been reported that higher altitudes are associated with enhanced survival rates of COVID-19 patients ( 24 ) and decreased COVID-19 fatality ( 25 ). This study reported a 47.1% mortality rate in ICU-admitted COVID-19 patients in Zanjan, Iran, where the altitude is 1638 meters above sea level. Similar studies on the Iranian population in an identical health system, conducted on 204 and 133 ICU-admitted COVID-19 patients in Tehran, Iran (at a lower elevation of 1200 m), indicated higher mortality rates (55.9% and 57.9%, respectively) ( 26 , 27 ). In a study of 184,875 ICU COVID-19 patients, the SOFA score outperformed the qSOFA score and SIRS criteria, with AUCs of 0.753, 0.607, and 0.589, respectively ( 28 ). A study involving 204 ICU-admitted COVID-19 patients examined the use of SOFA and APACHE II scores in predicting mortality. The area under the curve was 89.5% for SOFA and 73% for the APACHE II score, and SOFA showed a higher predictive value compared to the APACHE II score ( 27 ). Hence, we investigated the predictability of the SOFA system in ICU-admitted patients with COVID-19. The current study also demonstrated that deceased patients had significantly higher CO-RADS and SOFA scores. In another survey of 134 COVID-19 patients, the CO-RADS scores and SOFA scores in ICU patients were considerably higher than those in patients admitted to clinical wards ( 29 ). We investigated the biomarkers and clinical features of COVID-19 patients and their potential in predicting mortality using ROC analyses and odds ratios. All clinical records were recorded on ICU admission. AUC-ROC was the highest for LDH (> 437.5 U/L, AUC: 0.737, 95% CI: 0.690–0.783, OR: 3.717), followed by IL-6 (> 18.95 pg/mL, AUC: 0.761, 95% CI: 0.711–0.810, OR: 26.706) and SOFA score (> 3, AUC: 0.701, CI:0.652–0.751, OR: 4.221). A study involving 1,003 COVID-19 patients in France aimed to combine deep learning with CT scans, biological, and clinical data to predict the severity of COVID-19 patients. The findings indicated that AI-driven CT scans outperformed those evaluated by radiologists for several reasons. First, AI-driven CT scans had a higher AUC of 0.75 compared to 0.66 from radiologists' evaluations for predicting disease severity in a 135-patient validation cohort from Institut Gustave Roussy (IGR). Additionally, AI can integrate clinical features such as age and sex and capture some clinical information when data is missing in its assessments. Radiologists must read numerous cases and prioritize COVID-19 patients, while AI can help alleviate this burden by processing data more efficiently ( 30 ). Recent advancements in clinical practice, the emergence of new interventions and treatments, and the evolution of patient monitoring methods over the past 25 years have created a significant need to update the SOFA score ( 31 ). In a retrospective observational study involving 117 hospitalized patients with COVID-19, researchers sought to investigate the prognostic value of the SOFA score in these patients. The AUC for the diagnostic accuracy of the SOFA in predicting severe COVID-19 was 0.908, with a cutoff value of 2, a sensitivity of 85.20%, and a specificity of 80.40%. The SOFA score was assessed at the onset of the disease or upon admission ( 32 ). Another study of 320 patients with COVID-19 pneumonia who experienced severe respiratory distress employed ROC analysis to predict mortality, reporting an AUC of 0.883, a sensitivity of 74%, and a specificity of 80% at a cutoff point (SOFA scores = 5). They assessed the worst SOFA score recorded within 48 hours of the onset of severe respiratory distress ( 16 ). In another study on SOFA in various infectious states, the authors evaluated the worst SOFA score within 24 hours of ICU admission ( 33 ). The calculation time for the SOFA score varies among other studies. In our study, the retrospective design and the lack of sequential SOFA score recording precluded the possibility of determining the worst SOFA score over time; hence, we assessed the SOFA score recorded upon ICU admission. In addition to the differences in study populations and designs among studies, the lack of worst SOFA records may explain the differences in determining the optimum cut-off point for the SOFA score. IL-6 is recognized as a prototype cytokine that plays a significant role in host defense through its pleiotropic activities, which are released by various cell types. IL-6 stimulates immune responses and promotes acute-phase reactions as well as hematopoiesis ( 34 ). In our study, consistent with a previous study ( 35 ), we identified IL-6 as a significant predictor of mortality in COVID-19 patients, with an AUC of 0.761, CI: 0.711–0.811, specificity of 98.4%, and sensitivity of 65.5% at a cut-off value of 18.95. A study by Liu et al. indicated that monitoring IL-6 levels alone can reliably predict disease severity, although it was not linked to mortality ( 36 ). Additionally, another study showed that IL-6 is a more effective predictor of the need for mechanical ventilation than mortality ( 37 ). LDH is an intracellular enzyme that facilitates anaerobic glycolysis by converting pyruvate into lactate, which is routinely tested as a key indicator of inflammation ( 38 ). In a study on 203 COVID-19 patients, serum LDH level higher than 359.50 U/L predicted COVID-19 mortality with 93.8% sensitivity and 88.2% specificity, and LDH was identified as an independent predictor of COVID-19 mortality ( 39 ). In our study, which included ICU-admitted COVID-19 patients, the serum LDH level at the cut-off point (437.5 U/L) demonstrated 63.0% sensitivity and 68.6% specificity for predicting mortality. Pan et al. developed a machine learning-based model to predict mortality using eight factors: lymphocyte percentage, prothrombin time, LDH, total bilirubin, and eosinophil percentage, creatinine, neutrophil percentage, and albumin level. This model predicted 5-fold cross validation (AUC = 0.86) and the verification queue (AUC = 0.92). The best recognition performance was observed with an AUC of 0.86 in the 5-fold cross-validation training set, further improving to 0.92 in the validation cohort ( 40 ). Another prediction model incorporated key predictors, including age, chronic lung disease, CRP, D-dimer, NLR, creatinine, and total bilirubin. It showed intense discrimination in both the training (AUC = 0.912, 95% CI: 0.884–0.940) and validation cohorts (AUC = 0.922, 95% CI: 0.891–0.953), with good calibration ( 41 ). Previous studies have investigated the prognostic value of clinical parameters; however, there is limited research on integrating the predictive value of clinical, laboratory, and imaging findings to enhance COVID-19 mortality prediction in the intensive care setting. In the present study, models were designed incorporating clinical predictors and using ML algorithms. Eight models demonstrated an AUC higher than 0.800 in logistic regression and ML algorithms. Internal validation was performed using an integrated 10-fold stratified cross-validation with bootstrap. The six-factor model showed the best performance in the AdaBoost algorithm, with an AUC of 0.960 (95% CI: 0.941–0.974). The combination of cross-validation with bootstrapped confidence intervals ensured robust internal validation and minimized the risk of overestimating model performance. Incorporating both discrimination and calibration metrics, including AUC and Brier score, offered a more comprehensive evaluation of clinical predictive models. Despite internal validation results, our models lack external validation. Additionally, the predictive model should be interpreted cautiously due to the limitations of retrospective data, and further validation in larger prospective cohorts is necessary before clinical implementation. Validating these models in independent cohorts is essential to confirm their performance across diverse clinical settings and to support their broader clinical application. Limitations The retrospective nature of this study and the absence of a longitudinal cohort design imposed significant limitations on developing a prediction model. This may introduce potential selection and information biases that could affect the reliability of the results. Additionally, our prediction model was developed and tested in a single-center study, which raises the possibility of overfitting and limits its generalizability. Therefore, future prospective multicenter studies with external validation are needed to confirm and improve the predictive performance of our model. Due to the retrospective design, not all variables were collected for every patient, and missing data were observed in some variables. Moreover, the lack of a longitudinal cohort design precluded the possibility of follow-up on patients and evaluating temporal changes in clinical parameters to assess their dynamic prognostic value over time. The lack of longitudinal data limited the ability to infer the causality between the predictors and mortality outcomes. A prediction model development requires external validation (using independent datasets) and temporal validation to assess model performance over time. This study did not include external validation due to limited access to external datasets. As a result, our findings may not fully generalize to other patient populations or clinical settings. The incomplete patient records and the need for all the measurements required to calculate SOFA or CT scan scores. On the other hand, the limited prevalence of some underlying diseases prevented researchers from conducting more accurate statistical tests; therefore, it was impossible to investigate the effect of these underlying diseases on disease outcomes. It is recommended that further studies examine the impact of these diseases on COVID-19 outcomes as well as imaging findings, using a case-control design. Conclusion Clinical parameters, including IL-6, LDH, and SOFA, showed a good predictive value for mortality in ICU-admitted COVID-19 patients. Multi-organ assessment using the SOFA score showed a greater AUC for mortality predictability compared to the CT-based CO-RADS system in critical care. Machine learning models have enhanced the predictability of mortality. The six-factor AdaBoost model, which included age, IL-6, LDH, SOFA, CO-RADS, and ESR, demonstrated the best performance in predicting mortality risk, and internal validation of the model indicated good predictive performance. Machine learning models can serve as valuable prognostic tools for the early identification of critically ill patients, playing a crucial role in optimizing the allocation of medical resources, guiding patient triage, informing personalized treatment decision-making, and monitoring the progression of COVID-19. Declarations Ethics approval The study was approved by the Research Ethics Committee of Zanjan University of Medical Sciences, Zanjan, Iran (IR.ZUMS.REC.1401.204). This study was conducted in full accordance with the ethical principles outlined in the Declaration of Helsinki (World Medical Association, 2013) . Informed consent to participate was obtained from all of the participants Clinical trial number Not applicable. Funding resources The authors did not receive any specific grant from funding agencies in the public, commercial, or not-for-profit sectors. Data availability statement Data supporting the findings in this study are immediately available upon reasonable request. Declaration of Competing Interest The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper. Consent to Publication Not applicable. Acknowledgment Not applicable. Author information Not applicable. References Jiang S, Shi Z, Shu Y, Song J, Gao GF, Tan W, et al. A distinct name is needed for the new coronavirus. Vol. 395, Lancet (London, England). England; 2020. p. 949. Overview WHOR. COVID-19 Epidemiological Update. 2024;(November). Contreras S, Priesemann V. Risking further COVID-19 waves despite vaccination. Lancet Infect Dis. 2021 Jun;21(6):745–6. Wang D, Hu B, Hu C, Zhu F, Liu X, Zhang J, et al. Clinical Characteristics of 138 Hospitalized Patients with 2019 Novel Coronavirus-Infected Pneumonia in Wuhan, China. JAMA - J Am Med Assoc. 2020;323(11):1061–9. Grasselli G, Zangrillo A, Zanella A, Antonelli M, Cabrini L, Castelli A, et al. Baseline Characteristics and Outcomes of 1591 Patients Infected with SARS-CoV-2 Admitted to ICUs of the Lombardy Region, Italy. JAMA - J Am Med Assoc. 2020;323(16):1574–81. Dahai Zhao, Feifei Yao, Lijie Wang, Ling Zheng, Yongjun Gao, Jun Ye, et al. A Comparative Study on the Clinical Features of Coronavirus 2019 (COVID-19) Pneumonia With Other Pneumonias. Clin Infect Dis. 2020;71(15):756–61. Meyer NJ, Gattinoni L, Calfee CS. Acute respiratory distress syndrome. Lancet (London, England). 2021 Aug;398(10300):622–37. Hsu CY, Lai CC, Yeh YP, Chang-Chuan C, Chen HH. Progression from Pneumonia to ARDS as a Predictor for Fatal COVID-19. J Infect Public Health. 2021 Apr;14(4):504–7. Feng Z, Yu Q, Yao S, Luo L, Zhou W, Mao X, et al. Early prediction of disease progression in COVID-19 pneumonia patients with chest CT and clinical characteristics. Nat Commun. 2020;11(1):1–9. Herpe G, Lederlin M, Naudin M, Ohana M, Chaumoitre K, Gregory J, et al. Efficacy of Chest CT for COVID-19 Pneumonia Diagnosis in France. Radiology. 2021 Feb;298(2):E81–7. Shah V, Keniya R, Shridharani A, Punjabi M, Shah J, Mehendale N. Diagnosis of COVID-19 using CT scan images and deep learning techniques. Emerg Radiol. 2021;28(3):497–505. Prokop M, van Everdingen W, van Rees Vellinga T, Quarles van Ufford H, Stöger L, Beenen L, et al. CO-RADS: A Categorical CT Assessment Scheme for Patients Suspected of Having COVID-19-Definition and Evaluation. Radiology. 2020 Aug;296(2):E97–104. Chen SJ, Chao TF, Chiang MC, Kuo SC, Chen LY, Yin T, et al. Prediction of patient outcome from Acinetobacter baumannii bacteremia with sequential organ failure assessment (SOFA) and acute physiology and chronic health evaluation (APACHE) II scores. Intern Med. 2011;50(8):871–7. Ai Ghozy AR, El Hallage M, Khalid M, Abdalla K. The quick sequential organ failure assessment score for predicting outcome in patients with sepsis and evidence of multiorgan failure at the time of emergency department presentation. Crit Care Shock. 2022;25(3):143–56. Puhr-Westerheide D, Reich J, Sabel BO, Kunz WG, Fabritius MP, Reidler P, et al. Article sequential organ failure assessment outperforms quantitative chest ct imaging parameters for mortality prediction in covid-19 ards. Diagnostics. 2022;12(1):1–12. Fayed M, Patel N, Angappan S, Nowak K, Vasconcelos F, Penning DH, et al. Sequential Organ Failure Assessment ( SOFA ) Score and Mortality Prediction in Patients With Severe Respiratory Distress Secondary to COVID-. 2022;14(7). Lambden S, Laterre PF, Levy MM, Francois B. The SOFA score - Development, utility and challenges of accurate assessment in clinical trials. Crit Care. 2019;23(1):1–9. Penha D, Pinto EG, Matos F, Hochhegger B, Monaghan C, Taborda-Barata L, et al. CO-RADS: Coronavirus classification review. J Clin Imaging Sci. 2021;11(1):1–10. Emami A, Javanmardi F, Pirbonyeh N, Akbari A. Prevalence of Underlying Diseases in Hospitalized Patients with COVID-19: a Systematic Review and Meta-Analysis. Arch Acad Emerg Med. 2020;8(1):e35. Mehraeen E, Karimi A, Barzegary A, Vahedi F, Afsahi AM, Dadrase O, et al. Since January 2020 Elsevier has created Predictors of mortality in patients with COVID-19–a systematic review. Eur J Integr Med. 2020;40(January):101226. Banerjee A, Pasea L, Harris S, Gonzalez-Izquierdo A, Torralbo A, Shallcross L, et al. Estimating excess 1-year mortality associated with the COVID-19 pandemic according to underlying conditions and age: a population-based cohort study. Lancet. 2020;395(10238):1715–25. Figliozzi S, Masci PG, Ahmadi N, Tondi L, Koutli E, Aimo A, et al. Predictors of adverse prognosis in COVID-19: A systematic review and meta-analysis. Eur J Clin Invest. 2020;50(10):1–15. Bonanad C, García-blas S. Since January 2020 Elsevier has created a COVID-19 resource centre with free information in English and Mandarin on the novel coronavirus COVID- 19 . The COVID-19 resource centre is hosted on Elsevier Connect , the company ’ s public news and information. Jamda. 2020;(January):915–8. Simbaña-Rivera K, Morocho Jaramillo PR, Velastegui Silva J V., Gómez-Barreno L, Ventimilla Campoverde AB, Novillo Cevallos JF, et al. High-altitude is associated with better short-term survival in critically ill COVID-19 patients admitted to the ICU. PLoS One. 2022;17(3 March):1–20. Bridgman C, Gerken J, Vincent J, Brooks AE, Zapata I. Revisiting the COVID ‑ 19 fatality rate and altitude association through a comprehensive analysis. Sci Rep [Internet]. 2022;1–10. Available from: https://doi.org/10.1038/s41598-022-21787-z Vahedi A, Tabasi F, Monjazebi F, Hashemian SMR, Tabarsi P, Farzanegan B, et al. Clinical features and outcomes of ICU patients with COVID-19 infection in Tehran, Iran: A single-centered retrospective cohort study. Tanaffos. 2020;19(4):300–11. Beigmohammadi MT, Amoozadeh L, Rezaei Motlagh F, Rahimi M, Maghsoudloo M, Jafarnejad B, et al. Mortality Predictive Value of APACHE II and SOFA Scores in COVID-19 Patients in the Intensive Care Unit. Can Respir J. 2022;2022. Raith EP, Udy AA, Bailey M, McGloughlin S, MacIsaac C, Bellomo R, et al. Prognostic accuracy of the SOFA score, SIRS criteria, and qSOFA score for in-hospital mortality among adults with suspected infection admitted to the intensive care unit. JAMA - J Am Med Assoc. 2017;317(3):290–300. El Aidaoui K, Haoudar A, Khalis M, Kantri A, Ziati J, El Ghanmi A, et al. Predictors of Severity in Covid-19 Patients in Casablanca, Morocco. Cureus. 2020;12(9). Lassau N, Ammari S, Chouzenoux E, Gortais H, Herent P, Devilder M, et al. Integrating deep learning CT-scan model, biological and clinical variables to predict severity of COVID-19 patients. Nat Commun. 2021;12(1):1–11. Moreno R, Rhodes A, Piquilloud L, Hernandez G, Takala J, Gershengorn HB, et al. The Sequential Organ Failure Assessment ( SOFA ) Score : has the time come for an update ? Crit Care. 2023;1–5. Yang Z, Hu Q, Huang F, Xiong S, Sun Y. The prognostic value of the SOFA score in patients with COVID-19. 2021;32(June):1–9. Pawar RD, Shih JA, Balaji L, Grossestreuer A V., Patel P V., Hansen CK, et al. Variation in SOFA (Sequential Organ Failure Assessment) Score Performance in Different Infectious States. J Intensive Care Med. 2021;36(10):1217–22. Wang X, Tang G, Liu Y, Zhang L, Chen B, Han Y, et al. The role of IL-6 in coronavirus, especially in COVID-19. Front Pharmacol. 2022;13(November):1–12. Vélez-Páez JL, Baldeón-Rojas L, Cañadas Herrera C, Montalvo MP, Jara FE, Aguayo-Moscoso S, et al. Receiver operating characteristic (ROC) to determine cut-off points of clinical and biomolecular markers to discriminate mortality in severe COVID-19 living at high altitude. BMC Pulm Med [Internet]. 2023;23(1):1–10. Available from: https://doi.org/10.1186/s12890-023-02691-2 Liu X, Wang H, Shi S, Xiao J. Association between IL-6 and severe disease and mortality in COVID-19 disease: a systematic review and meta-analysis. Postgrad Med J. 2021;98(1165):871–9. Herold T, Jurinovic V, Arnreich C, Lipworth BJ, Hellmuth JC, von Bergwelt-Baildon M, et al. Elevated levels of IL-6 and CRP predict the need for mechanical ventilation in COVID-19. J Allergy Clin Immunol [Internet]. 2020;146(1):128-136.e4. Available from: https://doi.org/10.1016/j.jaci.2020.05.008 Komolafe O, Sp P, Br D, Ks G. for the diagnosis of pancreatic necrosis ( Review ). 2017; Li C, Ye J, Chen Q, Hu W, Wang L, Fan Y, et al. Elevated LDH. Definitions. 2020;12(15):15670–81. Pan P, Li Y, Xiao Y, Han B, Su L, Su M, et al. Prognostic assessment of COVID-19 in the intensive care unit by machine learning methods: Model development and validation. J Med Internet Res. 2020;22(11). Chen B, Gu HQ, Liu (刘艺) Y, Zhang G, Yang H, Hu H, et al. A model to predict the risk of mortality in severely ill COVID-19 patients. Comput Struct Biotechnol J [Internet]. 2021;19:1694–700. Available from: https://doi.org/10.1016/j.csbj.2021.03.012 Additional Declarations No competing interests reported. Cite Share Download PDF Status: Posted Version 1 posted You are reading this latest preprint version Research Square lets you share your work early, gain feedback from the community, and start making changes to your manuscript prior to peer review in a journal. As a division of Research Square Company, we’re committed to making research communication faster, fairer, and more useful. We do this by developing innovative software and high quality services for the global research community. Our growing team is made up of researchers and industry professionals working together to solve the most critical problems facing scientific publishing. Also discoverable on Platform About Our Team In Review Editorial Policies Advisory Board Help Center Resources Author Services Accessibility API Access RSS feed Manage Cookie Preferences © Research Square 2026 | ISSN 2693-5015 (online) Privacy Policy Terms of Service Do Not Sell My Personal Information {"props":{"pageProps":{"initialData":{"identity":"rs-8173098","acceptedTermsAndConditions":true,"allowDirectSubmit":true,"archivedVersions":[],"articleType":"Research Article","associatedPublications":[],"authors":[{"id":553716671,"identity":"522afbd6-f5ab-4e55-91aa-7bb6f86b7c09","order_by":0,"name":"Ali Heidarnejad Maleki","email":"","orcid":"","institution":"Zanjan University of Medical Sciences","correspondingAuthor":false,"prefix":"","firstName":"Ali","middleName":"Heidarnejad","lastName":"Maleki","suffix":""},{"id":553716672,"identity":"d93df3ee-6904-4bfc-846a-ee5bc6f1f450","order_by":1,"name":"Zeinab Khanali","email":"","orcid":"","institution":"Zanjan University of Medical Sciences","correspondingAuthor":false,"prefix":"","firstName":"Zeinab","middleName":"","lastName":"Khanali","suffix":""},{"id":553716673,"identity":"f60f620e-46f6-4a86-a3a9-5c742a4786b8","order_by":2,"name":"Ali Khayami","email":"","orcid":"","institution":"Zanjan University of Medical Sciences","correspondingAuthor":false,"prefix":"","firstName":"Ali","middleName":"","lastName":"Khayami","suffix":""},{"id":553716674,"identity":"2934b304-a8cd-4e3f-9a34-f7d6838a63af","order_by":3,"name":"Samad Ghodarti","email":"data:image/png;base64,iVBORw0KGgoAAAANSUhEUgAAAZAAAAAyAQMAAABI0h/eAAAABlBMVEX///8AAABVwtN+AAAACXBIWXMAAA7EAAAOxAGVKw4bAAAAyElEQVRIiWNgGAWjYLACHgYGxn4GBjYStcxsIFnLhgPEapFvP5344Q3DNtnNN5KfPfhQwSDPL3YAvxaDM7mbJecw3DbediPN3HDGGQbDmbMTCGhhyN0gzcNwO3HbjQQzad42hgSD2wS0yPe/3fwbpGXzjPRvxGlhuJG7DWzLBokcIm0xuPF2myXILzPOvCmTnHFGgrBf5PtzN994w3Bbtr89fZvEhwobeX5pQg4DAcZ/QEIArFKCCOVwwH+AFNWjYBSMglEwkgAAKxRFvWAULFwAAAAASUVORK5CYII=","orcid":"","institution":"Zanjan University of Medical Sciences","correspondingAuthor":true,"prefix":"","firstName":"Samad","middleName":"","lastName":"Ghodarti","suffix":""}],"badges":[],"createdAt":"2025-11-21 11:23:34","currentVersionCode":1,"declarations":"","doi":"10.21203/rs.3.rs-8173098/v1","doiUrl":"https://doi.org/10.21203/rs.3.rs-8173098/v1","draftVersion":[],"editorialEvents":[],"editorialNote":"","failedWorkflow":false,"files":[{"id":97299309,"identity":"d1c3409a-2d80-43b9-950f-2187e126531a","added_by":"auto","created_at":"2025-12-03 00:50:10","extension":"png","order_by":0,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":471049,"visible":true,"origin":"","legend":"","description":"","filename":"Figure1.png","url":"https://assets-eu.researchsquare.com/files/rs-8173098/v1/1ac07f9aead0d2cab70012db.png"},{"id":97299315,"identity":"1191cac9-9add-4b57-b43f-4d12da8d72c0","added_by":"auto","created_at":"2025-12-03 00:50:11","extension":"png","order_by":1,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":314290,"visible":true,"origin":"","legend":"","description":"","filename":"Figure2.png","url":"https://assets-eu.researchsquare.com/files/rs-8173098/v1/301c606144534edaced2680f.png"},{"id":97299311,"identity":"34cd74e2-39ac-4bd9-9902-c7d93f446a95","added_by":"auto","created_at":"2025-12-03 00:50:11","extension":"png","order_by":2,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":748664,"visible":true,"origin":"","legend":"","description":"","filename":"Figure3.png","url":"https://assets-eu.researchsquare.com/files/rs-8173098/v1/88e6c4b93f3406f74d6e9f13.png"},{"id":97369056,"identity":"a792908e-1eaa-4eaa-853c-1cff6030b252","added_by":"auto","created_at":"2025-12-03 16:23:34","extension":"docx","order_by":3,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":1054149,"visible":true,"origin":"","legend":"","description":"","filename":"Manuscript1.docx","url":"https://assets-eu.researchsquare.com/files/rs-8173098/v1/1483cb52a7c3923d41d3c23f.docx"},{"id":97369877,"identity":"bb894866-b976-4251-b2ee-ffd529ad338a","added_by":"auto","created_at":"2025-12-03 16:25:57","extension":"docx","order_by":4,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":17497,"visible":true,"origin":"","legend":"","description":"","filename":"Table1.docx","url":"https://assets-eu.researchsquare.com/files/rs-8173098/v1/f9591d7808b5196ffb3558b1.docx"},{"id":97369130,"identity":"b7ba8317-0a12-41e8-9d65-f0c8f5e5e7f7","added_by":"auto","created_at":"2025-12-03 16:23:45","extension":"docx","order_by":5,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":19677,"visible":true,"origin":"","legend":"","description":"","filename":"Table2.docx","url":"https://assets-eu.researchsquare.com/files/rs-8173098/v1/ef72915b4a6b8cd70e8b8ca6.docx"},{"id":97369849,"identity":"05319721-7f00-4a62-ac3f-915c4f735ad9","added_by":"auto","created_at":"2025-12-03 16:25:55","extension":"docx","order_by":6,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":19438,"visible":true,"origin":"","legend":"","description":"","filename":"Table3.docx","url":"https://assets-eu.researchsquare.com/files/rs-8173098/v1/fb9b4853af7ca4d3c1029d2c.docx"},{"id":97299320,"identity":"c4e397d8-e6a6-480b-86d5-e5e500b1a4eb","added_by":"auto","created_at":"2025-12-03 00:50:11","extension":"docx","order_by":7,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":15062,"visible":true,"origin":"","legend":"","description":"","filename":"Table4.docx","url":"https://assets-eu.researchsquare.com/files/rs-8173098/v1/436bf6e3cb9318d8ba128a7c.docx"},{"id":97369261,"identity":"4d072206-038f-4e27-a0e0-2c4683c243e7","added_by":"auto","created_at":"2025-12-03 16:24:02","extension":"docx","order_by":8,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":15139,"visible":true,"origin":"","legend":"","description":"","filename":"Table5.docx","url":"https://assets-eu.researchsquare.com/files/rs-8173098/v1/96b39e78169e3ebffe201823.docx"},{"id":97369356,"identity":"042f2e7f-fd6a-4eca-a57c-ebd3bc8488c6","added_by":"auto","created_at":"2025-12-03 16:24:33","extension":"json","order_by":9,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":7143,"visible":true,"origin":"","legend":"","description":"","filename":"3d11fe921eff4850b221cdfabd050983.json","url":"https://assets-eu.researchsquare.com/files/rs-8173098/v1/41da085944086127825e008b.json"},{"id":97369534,"identity":"569357ae-9bc5-4e0b-91d6-7da580ae34c3","added_by":"auto","created_at":"2025-12-03 16:25:06","extension":"xml","order_by":10,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":224099,"visible":true,"origin":"","legend":"","description":"","filename":"3d11fe921eff4850b221cdfabd0509831enriched.xml","url":"https://assets-eu.researchsquare.com/files/rs-8173098/v1/eb6c95ac3d6034468dc522f6.xml"},{"id":97368801,"identity":"f3439ba1-5cd2-4667-b2cb-1162893f95c8","added_by":"auto","created_at":"2025-12-03 16:22:57","extension":"png","order_by":11,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":471049,"visible":true,"origin":"","legend":"","description":"","filename":"Figure1.png","url":"https://assets-eu.researchsquare.com/files/rs-8173098/v1/cd30e676357b569d3af642b9.png"},{"id":97299335,"identity":"0230408b-9858-40f4-81e3-85cbb6aba02a","added_by":"auto","created_at":"2025-12-03 00:50:11","extension":"png","order_by":12,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":314290,"visible":true,"origin":"","legend":"","description":"","filename":"Figure2.png","url":"https://assets-eu.researchsquare.com/files/rs-8173098/v1/71a4f236f2443ce57978cda7.png"},{"id":97369374,"identity":"08624a79-3518-48d6-be2a-37e60cd2b9b6","added_by":"auto","created_at":"2025-12-03 16:24:36","extension":"png","order_by":13,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":748664,"visible":true,"origin":"","legend":"","description":"","filename":"Figure3.png","url":"https://assets-eu.researchsquare.com/files/rs-8173098/v1/6eb99e22ff30df25075b4d70.png"},{"id":97299331,"identity":"cb7ecbd0-3763-49da-a8bb-2c8530639231","added_by":"auto","created_at":"2025-12-03 00:50:11","extension":"jpeg","order_by":14,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":503920,"visible":true,"origin":"","legend":"","description":"","filename":"floatimage1.jpeg","url":"https://assets-eu.researchsquare.com/files/rs-8173098/v1/c8eb02bf30cca6870ff10d13.jpeg"},{"id":97299330,"identity":"1b6e1cf4-e4be-4140-a91a-50948a8d6f99","added_by":"auto","created_at":"2025-12-03 00:50:11","extension":"jpeg","order_by":15,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":535497,"visible":true,"origin":"","legend":"","description":"","filename":"floatimage2.jpeg","url":"https://assets-eu.researchsquare.com/files/rs-8173098/v1/8d45c21dd8ab5581a1ac45da.jpeg"},{"id":97299325,"identity":"606d2fdf-3a05-4487-93b0-776a441982b6","added_by":"auto","created_at":"2025-12-03 00:50:11","extension":"jpeg","order_by":16,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":753974,"visible":true,"origin":"","legend":"","description":"","filename":"floatimage3.jpeg","url":"https://assets-eu.researchsquare.com/files/rs-8173098/v1/cb357092d38600bc0829fdd3.jpeg"},{"id":97299323,"identity":"0af0cd16-b161-49f9-b7c2-969f687e75c7","added_by":"auto","created_at":"2025-12-03 00:50:11","extension":"png","order_by":17,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":79052,"visible":true,"origin":"","legend":"","description":"","filename":"OnlineFigure1.png","url":"https://assets-eu.researchsquare.com/files/rs-8173098/v1/53856858e3c4b15e89a912c6.png"},{"id":97299333,"identity":"356279bc-9bcd-4866-beb5-75e9b02ad317","added_by":"auto","created_at":"2025-12-03 00:50:11","extension":"png","order_by":18,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":98611,"visible":true,"origin":"","legend":"","description":"","filename":"OnlineFigure2.png","url":"https://assets-eu.researchsquare.com/files/rs-8173098/v1/f58fbef4975233c4f304858f.png"},{"id":97369166,"identity":"6a10aae0-2737-4c52-94ee-29acb99f820e","added_by":"auto","created_at":"2025-12-03 16:23:47","extension":"png","order_by":19,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":203996,"visible":true,"origin":"","legend":"","description":"","filename":"OnlineFigure3.png","url":"https://assets-eu.researchsquare.com/files/rs-8173098/v1/df2e8167743a4c762e7bd9db.png"},{"id":97369938,"identity":"88f68c50-5233-49ba-9337-7c163111cd2a","added_by":"auto","created_at":"2025-12-03 16:26:06","extension":"png","order_by":20,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":115117,"visible":true,"origin":"","legend":"","description":"","filename":"Onlinefloatimage1.png","url":"https://assets-eu.researchsquare.com/files/rs-8173098/v1/9afc121d7990db74372e7ac6.png"},{"id":97299334,"identity":"5e54c79b-18de-4bba-b61a-a47df9e8c672","added_by":"auto","created_at":"2025-12-03 00:50:11","extension":"png","order_by":21,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":112386,"visible":true,"origin":"","legend":"","description":"","filename":"Onlinefloatimage2.png","url":"https://assets-eu.researchsquare.com/files/rs-8173098/v1/1dcf478eb632b10293b0b75d.png"},{"id":97299332,"identity":"31974f86-52df-4dd0-94c7-6ac4db41353c","added_by":"auto","created_at":"2025-12-03 00:50:11","extension":"png","order_by":22,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":150871,"visible":true,"origin":"","legend":"","description":"","filename":"Onlinefloatimage3.png","url":"https://assets-eu.researchsquare.com/files/rs-8173098/v1/19952776ae52258b9a676acb.png"},{"id":97299321,"identity":"5ebff1d3-84da-4a30-a5f4-22c169396d58","added_by":"auto","created_at":"2025-12-03 00:50:11","extension":"xml","order_by":23,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":224386,"visible":true,"origin":"","legend":"","description":"","filename":"3d11fe921eff4850b221cdfabd0509831structuring.xml","url":"https://assets-eu.researchsquare.com/files/rs-8173098/v1/d72ec437103f67e232afdf3b.xml"},{"id":97369259,"identity":"e837bd0d-5ebb-4973-8f78-f778afbed73a","added_by":"auto","created_at":"2025-12-03 16:24:02","extension":"html","order_by":24,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":233832,"visible":true,"origin":"","legend":"","description":"","filename":"earlyproof.html","url":"https://assets-eu.researchsquare.com/files/rs-8173098/v1/6db6be7220bda26766e4c722.html"},{"id":97299310,"identity":"2d232bb6-e676-409a-97d1-c9499c02d440","added_by":"auto","created_at":"2025-12-03 00:50:11","extension":"png","order_by":1,"title":"Figure 1","display":"","copyAsset":false,"role":"figure","size":471049,"visible":true,"origin":"","legend":"\u003cp\u003eROC curves for the parameters associated with mortality in the bivariate analysis. (A) Age, (B) CRP, (C) IL-6, (D) LDH, (E) WBC, (F) Total bilirubin, (G) PaO2, (H) PaO\u003csub\u003e2\u003c/sub\u003e/FiO\u003csub\u003e2\u003c/sub\u003e (OI), (I) SOFA index, (J) CO-RADS index. The area under the curve (AUC), 95% CI, and cut-points of ROC curves are recorded in Table 3.\u003c/p\u003e","description":"","filename":"Figure1.png","url":"https://assets-eu.researchsquare.com/files/rs-8173098/v1/68cf0372489c16f430756ae4.png"},{"id":97369708,"identity":"23b3da02-d5a9-41c6-974f-f6cbddf80ae5","added_by":"auto","created_at":"2025-12-03 16:25:36","extension":"png","order_by":2,"title":"Figure 2","display":"","copyAsset":false,"role":"figure","size":314290,"visible":true,"origin":"","legend":"\u003cp\u003eROC curves for the predictive models (A) IL-6 + LDH, (B) Age + IL-6 + LDH, (C) Age + IL-6 + SOFA, (D) IL-6 + LDH + SOFA, (E) Age + IL-6 + SOFA + CO-RADS, (F) Age + IL-6 + LDH + SOFA, (G) Age + IL-6 + LDH + SOFA + CO-RADS, (H) Age + IL-6 + LDH + SOFA + CO-RADS + ESRin logistic regression (LR), Gradient-boosted decision trees (GBDT), AdaBoost, XGBoost, and CatBoost models.\u003c/p\u003e","description":"","filename":"Figure2.png","url":"https://assets-eu.researchsquare.com/files/rs-8173098/v1/1648100d8357d034e7c09555.png"},{"id":97299313,"identity":"ad8e26d4-5869-4d04-a93b-32edb953a9ec","added_by":"auto","created_at":"2025-12-03 00:50:11","extension":"png","order_by":3,"title":"Figure 3","display":"","copyAsset":false,"role":"figure","size":748664,"visible":true,"origin":"","legend":"\u003cp\u003eThe calibration curve demonstrates the agreement between the predicted risk and the observed risk for each model in the algorithms. The close alignment of the models predicted curve with the diagonal line indicates that the predicted probabilities are in good agreement with the actual outcomes. (A) IL-6 + LDH, (B) Age + IL-6 + LDH, (C) Age + IL-6 + SOFA, (D) IL-6 + LDH + SOFA, (E) Age + IL-6 + SOFA + CO-RADS, (F) Age + IL-6 + LDH + SOFA, (G) Age + IL-6 + LDH + SOFA + CO-RADS, (H) Age + IL-6 + LDH + SOFA + CO-RADS + ESR.\u003c/p\u003e","description":"","filename":"Figure3.png","url":"https://assets-eu.researchsquare.com/files/rs-8173098/v1/6c93e985e2337652dc31d03d.png"},{"id":100129300,"identity":"00b9eb07-a38f-45a0-8e8f-df875c60c4fd","added_by":"auto","created_at":"2026-01-13 09:55:27","extension":"pdf","order_by":0,"title":"","display":"","copyAsset":false,"role":"manuscript-pdf","size":2336550,"visible":true,"origin":"","legend":"","description":"","filename":"manuscript.pdf","url":"https://assets-eu.researchsquare.com/files/rs-8173098/v1/f39fe09d-5d98-43d1-a2d5-0ed634b25f55.pdf"}],"financialInterests":"No competing interests reported.","formattedTitle":"Machine Learning Prediction Models for COVID-19 ICU Mortality: Model Development and Validation","fulltext":[{"header":"Introduction","content":"\u003cp\u003eIn December 2019, a new coronavirus was identified as the cause of pneumonia in Wuhan, China. In February 2020, the World Health Organization (WHO) named the coronavirus disease 2019 (COVID-19) (\u003cspan citationid=\"CR1\" class=\"CitationRef\"\u003e1\u003c/span\u003e). According to the World Health Organization, as of October 13, 2024, more than 776\u0026nbsp;million verified cases and more than 7\u0026nbsp;million deaths have been reported worldwide (\u003cspan citationid=\"CR2\" class=\"CitationRef\"\u003e2\u003c/span\u003e). Even with coronavirus vaccinations, COVID-19 waves can still impact lives due to mutations that have occurred since the disease emerged (\u003cspan citationid=\"CR3\" class=\"CitationRef\"\u003e3\u003c/span\u003e). Studies indicate that 15\u0026ndash;30% of hospitalized COVID-19 patients require admission to an intensive care unit (ICU), and the mortality rate of ICU-admitted patients ranges from 30% to over 50%, depending on the healthcare system and patient demographics (\u003cspan citationid=\"CR4\" class=\"CitationRef\"\u003e4\u003c/span\u003e, \u003cspan citationid=\"CR5\" class=\"CitationRef\"\u003e5\u003c/span\u003e). Although the mortality rate of COVID-19 is as low as 3%, its rapid transmission resulted in a considerable number of deaths (\u003cspan citationid=\"CR6\" class=\"CitationRef\"\u003e6\u003c/span\u003e).\u003c/p\u003e\u003cp\u003eReverse transcriptase polymerase chain reaction (RT-PCR) is the gold standard diagnostic test for COVID-19. Following these studies, CT-scan signs of pulmonary involvement were incorporated into the diagnostic protocol for this disease in China. In addition, ARDS can be a complication of COVID-19, leading to disease progression and increased mortality (\u003cspan citationid=\"CR7\" class=\"CitationRef\"\u003e7\u003c/span\u003e, \u003cspan citationid=\"CR8\" class=\"CitationRef\"\u003e8\u003c/span\u003e). The risk stratification was often based on lung involvement, making CT scans valuable diagnostic tools during the pandemic (\u003cspan citationid=\"CR9\" class=\"CitationRef\"\u003e9\u003c/span\u003e, \u003cspan citationid=\"CR10\" class=\"CitationRef\"\u003e10\u003c/span\u003e). A Lung CT scan helps diagnose COVID-19 and can be crucial in evaluating the response to treatment and disease progression (\u003cspan citationid=\"CR11\" class=\"CitationRef\"\u003e11\u003c/span\u003e). The COVID-19 Reporting and Data System (CO-RADS), introduced by the Dutch Radiological Society, categorizes pulmonary involvement in COVID-19 based on non-enhanced chest CT scans. It effectively predicted the virus in patients with moderate to severe symptoms (\u003cspan citationid=\"CR12\" class=\"CitationRef\"\u003e12\u003c/span\u003e). The significant advantage of this classification is the ease of use.\u003c/p\u003e\u003cp\u003eThere was a need for disease severity scoring systems, such as SOFA and APACHE, to improve patient outcomes and optimize resource use. The Sequential Organ Failure Assessment (SOFA) is a crucial tool for evaluating multi-organ failure and is recognized as a predictive factor for outcomes in various diseases (\u003cspan citationid=\"CR13\" class=\"CitationRef\"\u003e13\u003c/span\u003e, \u003cspan citationid=\"CR14\" class=\"CitationRef\"\u003e14\u003c/span\u003e). SOFA scoring is based on assessing the function of six organs: blood circulation, respiratory system, liver, kidney, central nervous system (CNS), and coagulation system. The score assigned to each organ can range between 3 and 8. SOFA is a quick and accessible tool for rapidly evaluating hospitalized patients in acute conditions (\u003cspan citationid=\"CR13\" class=\"CitationRef\"\u003e13\u003c/span\u003e). In the present study, biomarkers were also investigated, including serum levels of interleukin 6 (IL-6), lactate dehydrogenase (LDH), sedimentation rate (ESR), C-reactive protein (CRP), white blood cell (WBC) count, lymphocyte count, platelet count, total bilirubin, and creatinine. As initial evaluation and treatment response surveillance in COVID-19 patients can prevent mortality and complications, identifying predictors of disease severity is crucial in these patients. Considering the ease of use of SOFA and the availability of CT scans as a diagnostic tool for COVID-19, these two methods are hypothetically helpful in evaluating the risk of mortality and COVID-19 severity.\u003c/p\u003e\u003cp\u003eRecent studies have investigated the correlation between chest CT findings, SOFA scores, and mortality in COVID-19 patients: A chest CT scan alone is insufficient to predict mortality in COVID-19 ARDS patients, and it is advisable to utilize the SOFA score for multi-organ assessment (\u003cspan citationid=\"CR15\" class=\"CitationRef\"\u003e15\u003c/span\u003e). The SOFA scoring system, along with imaging and other clinical findings, was recommended for predicting mortality in patients with COVID-19 pneumonia (\u003cspan citationid=\"CR16\" class=\"CitationRef\"\u003e16\u003c/span\u003e). This study aimed to investigate the predictive properties of various clinical parameters. While existing research has explored the prognostic value of clinical parameters, there is a limited understanding of integrating the predictive value of clinical, laboratory, and imaging findings to enhance COVID-19 mortality prediction in the intensive care setting. We designed ML-based models to develop mortality prediction performance and conducted internal validation for each model. The designed prediction models incorporate biomarkers and clinical findings, and depending on the availability of clinical data, these models may serve as valuable prognostic tools for the early identification of critically ill patients. This plays a crucial role in optimizing the allocation of medical resources, guiding patient triage, personalized treatment decision-making, and monitoring the progression of COVID-19.\u003c/p\u003e"},{"header":"Methods","content":"\u003cp\u003e The current study was a retrospective cohort study conducted among patients with COVID-19 who were admitted to the intensive care unit (ICU) department of a tertiary hospital in Zanjan, Iran, from March to November 2020. The study population was selected using census sampling.\u003c/p\u003e\u003cp\u003eThe inclusion criteria were an age of 18 years or older and a documented diagnosis of COVID-19, as confirmed by RT-PCR or CT scan. The exclusion criteria were pregnancy and incomplete patient files.\u003c/p\u003e\u003cp\u003eData were collected from patients' records and imported into a checklist. Data included demographic data (age and gender), past medical history (hypertention, diabetes, ischemic heart diseases (IHD), chronic obstructive pulmonary diseases (COPD), cancer, congestive heart failure (CHF), cerebrovascular accidents (CVA), hyperlipidemia (HLP), chronic kidney diseases (CKD)), vital signs (heart rate, respiratory rate, temperature, systolic and diastolic blood pressure, mean arterial pressure), laboratory findings (serum LDH, CRP, ESR, IL-6, WBC count, lymphocytes count, serum creatinine, total bilirubin, platelet count (Plt), and arterial oxygen pressure [PaO\u003csub\u003e2\u003c/sub\u003e]), spirometry findings (fraction of inspired oxygen [FiO\u003csub\u003e2\u003c/sub\u003e]), Oxygenation index (OI: PaO\u003csub\u003e2\u003c/sub\u003e/FiO\u003csub\u003e2\u003c/sub\u003e), consciousness level based on the Glasgow Coma Scale (GCS). The SOFA score was calculated from the patient records. The CO-RADS was evaluated based on the archived CT scans of the included patients.\u003c/p\u003e\u003cdiv id=\"Sec3\" class=\"Section2\"\u003e\u003ch2\u003eStudy instruments\u003c/h2\u003e\u003cp\u003eThe SOFA score was calculated based on the evaluation of the function of six organs (\u003cspan citationid=\"CR17\" class=\"CitationRef\"\u003e17\u003c/span\u003e). In each organ evaluation, normal function is assigned a score of zero, and the SOFA score increases with an increased level of dysfunction in the organ. The criteria for evaluating the respiratory system are PaO\u003csub\u003e2\u003c/sub\u003e/FiO\u003csub\u003e2\u003c/sub\u003e and the need for respiratory support. Based on these criteria, a PaO\u003csub\u003e2\u003c/sub\u003e/FiO\u003csub\u003e2\u003c/sub\u003e above 400 is scored as zero, a PaO\u003csub\u003e2\u003c/sub\u003e/FiO\u003csub\u003e2\u003c/sub\u003e between 301 and 400 is scored as one, a PaO\u003csub\u003e2\u003c/sub\u003e/FiO\u003csub\u003e2\u003c/sub\u003e between 201 and 300 is scored as two, a PaO\u003csub\u003e2\u003c/sub\u003e/FiO\u003csub\u003e2\u003c/sub\u003e between 101 and 200 with respiratory support is scored as three, and a PaO\u003csub\u003e2\u003c/sub\u003e/FiO\u003csub\u003e2\u003c/sub\u003e below 100 with respiratory backing is scored as four.\u003c/p\u003e\u003cp\u003ePlatelet count above 150 (1000/\u0026micro;l) is scored zero, while platelet counts of 101\u0026ndash;150 (1000/\u0026micro;l), 51\u0026ndash;100 (1000/\u0026micro;l), 21\u0026ndash;50 (1000/\u0026micro;l), and equal or lower than 20 (1000/\u0026micro;l) are scored, one, two, three, and four, respectively. Serum bilirubin below 1.2 mg/dL is scored zero, while serum bilirubin levels 1.2\u0026ndash;1.9 mg/dL, 2.0-5.9 mg/dL, 6.0-11.9 mg/dL, and above 12 mg/dL are scored one, two, three, and four, respectively. A mean arterial pressure (MAP) greater than or equal to 70 mmHg is scored zero. In comparison, MAP below 70 mmHg, need for 5 mg dopamine or any dose of dobutamine, need for more than 5 mg dopamine or epinephrine or norepinephrine less than or equal to 0.1, need for more than 15 mg dopamine or epinephrine or norepinephrine less than or equal to 0.1 are scored, one, two, three, and four, respectively.\u003c/p\u003e\u003cp\u003eSerum creatinine levels below 1.2 mg/dL are scored as zero. In contrast, serum creatinine levels of 1.2\u0026ndash;1.9 mg/dL, 2.0-3.4 mg/dL, 3.5\u0026ndash;4.9 mg/dL, or urine output less than 500 mL/d, serum creatinine levels more than 5.0 mg/dL, or urine output less than 200 mL are scored as one, two, three, and four, respectively. A GCS of 15 is scored zero, while GCS scores of 13\u0026ndash;14, 10\u0026ndash;12, 6\u0026ndash;9, and less than six are scored one, two, three, and four, respectively (\u003cspan citationid=\"CR17\" class=\"CitationRef\"\u003e17\u003c/span\u003e).\u003c/p\u003e\u003cp\u003eCT interpretations and the CO-RADS scoring were performed using the CO-RADS criteria by a qualified board-certified radiologist. The CO-RADS score was zero if the CT scan quality prevented the radiologist from categorizing any scores (\u003cspan citationid=\"CR18\" class=\"CitationRef\"\u003e18\u003c/span\u003e). A score of one is given for a routine CT scan or findings attributed to non-infectious disorders. Score two is given to a CT scan with conclusions related to infectious diseases but not COVID-19 (bronchitis, bronchiolitis, bronchopneumonia, centrilobular GGO, or pulmonary abscess). Score three is given to CT scan with unclear findings for COVID-19 (perihilar GGO, homogenous and widespread GGO, GGO related to interlobular thickening, and pneumonia patterns in case of the absence of other COVID-19 findings. A score of four is given to a CT scan with suspicious COVID-19 findings that may overlap with those of other viral pneumonias. A score of 5 is given to a CT scan with highly suggestive findings of COVID-19 (\u003cspan citationid=\"CR18\" class=\"CitationRef\"\u003e18\u003c/span\u003e).\u003c/p\u003e\u003c/div\u003e\u003cdiv id=\"Sec4\"\u003e\n \u003ch2\u003eStatistical analysis\u003c/h2\u003e\n \u003cp\u003eInitial statistical analyses were performed using IBM SPSS Statistics version 26 (IBM Corp., Armonk, NY, USA). The results of these analyses are presented in Tables\u0026nbsp;1–3 and Fig.\u0026nbsp;1. Advanced modeling and machine learning analyses were conducted using Python version 3.10 (Python Software Foundation, Wilmington, DE, USA), incorporating packages such as scikit-learn (sklearn), seaborn, matplotlib, pandas, and NumPy. The results of the Python-based analyses are presented in Tables\u0026nbsp;4 and 5, as well as Figs.\u0026nbsp;2 and 3.\u003c/p\u003e\n \u003cdiv\u003e\n \u003ctable id=\"Tab1\" border=\"1\"\u003e\n \u003ccaption language=\"En\"\u003e\n \u003cdiv\u003eTable 1\u003c/div\u003e\n \u003cdiv\u003e\n \u003cp\u003eDemographic and clinical characteristics\u003c/p\u003e\n \u003c/div\u003e\n \u003c/caption\u003e\n \u003cthead\u003e\n \u003ctr\u003e\n \u003cth align=\"left\" colspan=\"2\"\u003e\n \u003cp\u003eCharacteristic\u003c/p\u003e\n \u003c/th\u003e\n \u003cth align=\"left\"\u003e\n \u003cp\u003eOverall\u003c/p\u003e\n \u003cp\u003e(n = 426)\u003c/p\u003e\n \u003c/th\u003e\n \u003cth align=\"left\"\u003e\n \u003cp\u003eDischarged\u003c/p\u003e\n \u003cp\u003e(N = 226)\u003c/p\u003e\n \u003c/th\u003e\n \u003cth align=\"left\"\u003e\n \u003cp\u003eDeceased\u003c/p\u003e\n \u003cp\u003e(n = 200)\u003c/p\u003e\n \u003c/th\u003e\n \u003cth align=\"left\"\u003e\n \u003cp\u003ep-value\u003c/p\u003e\n \u003c/th\u003e\n \u003c/tr\u003e\n \u003c/thead\u003e\n \u003ctbody\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\" colspan=\"2\"\u003e\n \u003cp\u003eAge (y), mean (SD) years\u003csup\u003e1\u003c/sup\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e61.51 (16.90)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e57.57 (18.28)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e65.96 (13.96)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e\u0026lt; 0.001*\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\" colspan=\"2\"\u003e\n \u003cp\u003eGender, male, no (%)\u003csup\u003e2\u003c/sup\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e224 (52.60%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e121 (53.53%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e103 (51.50%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e0.714\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\" colspan=\"2\"\u003e\n \u003cp\u003eComorbidities, no (%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\u0026nbsp;\u003c/td\u003e\n \u003ctd align=\"left\"\u003e\u0026nbsp;\u003c/td\u003e\n \u003ctd align=\"left\"\u003e\u0026nbsp;\u003c/td\u003e\n \u003ctd align=\"left\"\u003e\u0026nbsp;\u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\" colspan=\"2\"\u003e\n \u003cp\u003eHypertension\u003csup\u003e2\u003c/sup\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e174 (40.8%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e76 (33.62%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e98 (49%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e0.001*\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\" colspan=\"2\"\u003e\n \u003cp\u003eDiabetes mellitus\u003csup\u003e2\u003c/sup\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e96 (22.53%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e42 (18.58%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e54 (27%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e0.038*\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\" colspan=\"2\"\u003e\n \u003cp\u003eIschemic heart disease\u003csup\u003e2\u003c/sup\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e43 (10.09%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e20 (8.85%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e23 (11.5%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e0.365\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\" colspan=\"2\"\u003e\n \u003cp\u003eChronic Obstructive Pulmonary Disease\u003csup\u003e2\u003c/sup\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e31 (7.28%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e15 (6.64%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e16 (8%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e0.589\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\" colspan=\"2\"\u003e\n \u003cp\u003eNeoplasia\u003csup\u003e2\u003c/sup\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e16 (3.76%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e5 (2.21%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e11 (5.5%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e0.075\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\" colspan=\"2\"\u003e\n \u003cp\u003eCongestive Heart Failure\u003csup\u003e2\u003c/sup\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e10 (2.35%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e7 (3.1%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e3 (1.5%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e0.277\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\" colspan=\"2\"\u003e\n \u003cp\u003eCerebrovascular accident\u003csup\u003e2\u003c/sup\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e13 (3.05%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e8 (3.54%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e5 (2.5%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e0.533\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\" colspan=\"2\"\u003e\n \u003cp\u003eHyperlipidemia\u003csup\u003e2\u003c/sup\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e17 (3.99%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e8 (3.54%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e9 (4.5%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e0.613\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\" colspan=\"2\"\u003e\n \u003cp\u003eChronic kidney disease\u003csup\u003e2\u003c/sup\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e13 (3.05%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e5 (2.21%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e8 (4%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e0.284\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\" colspan=\"2\"\u003e\n \u003cp\u003eMechanical ventilation, positive, no (%)\u003csup\u003e2\u003c/sup\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e48 (11.27%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e15 (6.64%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e33 (16.5%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e0.001*\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\" colspan=\"2\"\u003e\n \u003cp\u003eSARS-CoV-2 RT-PCR result, positive, no (%)\u003csup\u003e2\u003c/sup\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e292 (68.54%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e132 (58.40%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e160 (80%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e\u0026lt; 0.001*\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\" colspan=\"2\"\u003e\n \u003cp\u003eCO-RADS category, no (%)\u003csup\u003e1\u003c/sup\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e5.27 (1.12)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e5.03 (1.19)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e5.54 (0.96)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e\u0026lt; 0.001*\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\u0026nbsp;\u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e\u003cstrong\u003e0\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e0 (0%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e0 (0%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e0 (0%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\u0026nbsp;\u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\u0026nbsp;\u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e\u003cstrong\u003e1\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e0 (0%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e0 (0%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e0 (0%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\u0026nbsp;\u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\u0026nbsp;\u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e\u003cstrong\u003e2\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e0 (0%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e0 (0%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e0 (0%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\u0026nbsp;\u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\u0026nbsp;\u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e\u003cstrong\u003e3\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e46 (10.79%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e32 (14.15%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e14 (7.00%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\u0026nbsp;\u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\u0026nbsp;\u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e\u003cstrong\u003e4\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e86 (20.18%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e61 (26.99%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e25 (12.50%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\u0026nbsp;\u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\u0026nbsp;\u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e\u003cstrong\u003e5\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e2 (0.46%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e1 (0.44%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e1 (0.50%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\u0026nbsp;\u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\u0026nbsp;\u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e\u003cstrong\u003e6\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e292 (68.54%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e132 (58.40%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e160 (80.00%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\u0026nbsp;\u003c/td\u003e\n \u003c/tr\u003e\n \u003c/tbody\u003e\n \u003ctfoot\u003e\n \u003ctr\u003e\n \u003ctd colspan=\"6\"\u003eSD Standard deviation, no number, SOFA sequential organ failure assessment, CO-RADS COVID-19 Reporting and Data System, *significant differences (p‑value \u0026lt; 0.05), \u003csup\u003e1/\u003c/sup\u003e Mann-Whitney U test, \u003csup\u003e2/\u003c/sup\u003e Chi‑square test or exact Fisher test\u003c/td\u003e\n \u003c/tr\u003e\n \u003c/tfoot\u003e\n \u003c/table\u003e\n \u003c/div\u003e\n \u003cdiv\u003e\n \u003ctable id=\"Tab2\" border=\"1\"\u003e\n \u003ccaption language=\"En\"\u003e\n \u003cdiv\u003eTable 3\u003c/div\u003e\n \u003cdiv\u003e\n \u003cp\u003eVariables as potential predictors of mortality *PPV: Positive predictive value, **NPV: Negative predictive value\u003c/p\u003e\n \u003c/div\u003e\n \u003c/caption\u003e\n \u003cthead\u003e\n \u003ctr\u003e\n \u003cth align=\"left\"\u003e\u0026nbsp;\u003c/th\u003e\n \u003cth align=\"left\" colspan=\"3\"\u003e\n \u003cp\u003eROC\u003c/p\u003e\n \u003c/th\u003e\n \u003cth align=\"left\" colspan=\"3\"\u003e\n \u003cp\u003eRisk factor cutoff characterization\u003c/p\u003e\n \u003c/th\u003e\n \u003cth align=\"left\"\u003e\u0026nbsp;\u003c/th\u003e\n \u003cth align=\"left\"\u003e\u0026nbsp;\u003c/th\u003e\n \u003cth align=\"left\" colspan=\"4\"\u003e\n \u003cp\u003eCutoff univariate logistic regression\u003c/p\u003e\n \u003c/th\u003e\n \u003c/tr\u003e\n \u003c/thead\u003e\n \u003ctbody\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eVariables\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eAUC\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e95% CI\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eP value\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eCut-off\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eSensitivity\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eSpecificity\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003ePPV\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eNPV\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eValue\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eOR\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e95% CI\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eP value\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\" rowspan=\"2\"\u003e\n \u003cp\u003eAge\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\" rowspan=\"2\"\u003e\n \u003cp\u003e0.633\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\" rowspan=\"2\"\u003e\n \u003cp\u003e0.581–0.686\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\" rowspan=\"2\"\u003e\n \u003cp\u003e\u0026lt; 0.001\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\" rowspan=\"2\"\u003e\n \u003cp\u003e50\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\" rowspan=\"2\"\u003e\n \u003cp\u003e0.87\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\" rowspan=\"2\"\u003e\n \u003cp\u003e0.347\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\" rowspan=\"2\"\u003e\n \u003cp\u003e0.54\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\" rowspan=\"2\"\u003e\n \u003cp\u003e0.74\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e≤ 50\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eref\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\u0026nbsp;\u003c/td\u003e\n \u003ctd align=\"left\" rowspan=\"2\"\u003e\n \u003cp\u003e\u0026lt; 0.001\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e\u0026gt; 50\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e3.366\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e2.075–5.459\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\" rowspan=\"2\"\u003e\n \u003cp\u003eCRP\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\" rowspan=\"2\"\u003e\n \u003cp\u003e0.597\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\" rowspan=\"2\"\u003e\n \u003cp\u003e0.535–0.658\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\" rowspan=\"2\"\u003e\n \u003cp\u003e0.002\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\" rowspan=\"2\"\u003e\n \u003cp\u003e70.0 (mg/L)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\" rowspan=\"2\"\u003e\n \u003cp\u003e0.389\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\" rowspan=\"2\"\u003e\n \u003cp\u003e0.814\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\" rowspan=\"2\"\u003e\n \u003cp\u003e0.57\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\" rowspan=\"2\"\u003e\n \u003cp\u003e0.68\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e≤ 70 (mg/L)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eref\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\u0026nbsp;\u003c/td\u003e\n \u003ctd align=\"left\" rowspan=\"2\"\u003e\n \u003cp\u003e\u0026lt; 0.001\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e\u0026gt; 70\u003c/p\u003e\n \u003cp\u003e(mg/L)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e2.788\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e1.736–4.478\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\" rowspan=\"2\"\u003e\n \u003cp\u003eIL-6\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\" rowspan=\"2\"\u003e\n \u003cp\u003e0.761\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\" rowspan=\"2\"\u003e\n \u003cp\u003e0.711–0.810\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\" rowspan=\"2\"\u003e\n \u003cp\u003e\u0026lt; 0.001\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\" rowspan=\"2\"\u003e\n \u003cp\u003e18.95 (pg/mL)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\" rowspan=\"2\"\u003e\n \u003cp\u003e0.655\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\" rowspan=\"2\"\u003e\n \u003cp\u003e0.934\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\" rowspan=\"2\"\u003e\n \u003cp\u003e0.90\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\" rowspan=\"2\"\u003e\n \u003cp\u003e0.75\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e≤ 18.95 (pg/mL)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eref\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\u0026nbsp;\u003c/td\u003e\n \u003ctd align=\"left\" rowspan=\"2\"\u003e\n \u003cp\u003e\u0026lt; 0.001\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e\u0026gt; 18.95 (pg/mL)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e26.706\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e14.665–48.634\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\" rowspan=\"2\"\u003e\n \u003cp\u003eLDH\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\" rowspan=\"2\"\u003e\n \u003cp\u003e0.737\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\" rowspan=\"2\"\u003e\n \u003cp\u003e0.690–0.783\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\" rowspan=\"2\"\u003e\n \u003cp\u003e\u0026lt; 0.001\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\" rowspan=\"2\"\u003e\n \u003cp\u003e437.5 (U/L)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\" rowspan=\"2\"\u003e\n \u003cp\u003e0.630\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\" rowspan=\"2\"\u003e\n \u003cp\u003e0.686\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\" rowspan=\"2\"\u003e\n \u003cp\u003e0.64\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\" rowspan=\"2\"\u003e\n \u003cp\u003e0.68\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e≤ 437.5 (U/L)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eref\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\u0026nbsp;\u003c/td\u003e\n \u003ctd align=\"left\" rowspan=\"2\"\u003e\n \u003cp\u003e\u0026lt; 0.001\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e\u0026gt; 437.5 (U/L)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e3.717\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e2.488–5.554\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\" rowspan=\"2\"\u003e\n \u003cp\u003eWBC\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\" rowspan=\"2\"\u003e\n \u003cp\u003e0.555\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\" rowspan=\"2\"\u003e\n \u003cp\u003e0.500–0.610\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\" rowspan=\"2\"\u003e\n \u003cp\u003e0.049\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\" rowspan=\"2\"\u003e\n \u003cp\u003e10.15 (×10\u003csup\u003e9\u003c/sup\u003e/L)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\" rowspan=\"2\"\u003e\n \u003cp\u003e0.540\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\" rowspan=\"2\"\u003e\n \u003cp\u003e0.571\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\" rowspan=\"2\"\u003e\n \u003cp\u003e0.52\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\" rowspan=\"2\"\u003e\n \u003cp\u003e0.58\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e≤ 10.15 (×10\u003csup\u003e9\u003c/sup\u003e/L)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eref\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\u0026nbsp;\u003c/td\u003e\n \u003ctd align=\"left\" rowspan=\"2\"\u003e\n \u003cp\u003e0.022\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e\u0026gt; 10.15 (×10\u003csup\u003e9\u003c/sup\u003e/L)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e1.561\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e1.064–2.290\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\" rowspan=\"2\"\u003e\n \u003cp\u003eBilirubin (Total)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\" rowspan=\"2\"\u003e\n \u003cp\u003e0.582\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\" rowspan=\"2\"\u003e\n \u003cp\u003e0.528–0.636\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\" rowspan=\"2\"\u003e\n \u003cp\u003e0.003\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\" rowspan=\"2\"\u003e\n \u003cp\u003e1.15 (mg/dL)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\" rowspan=\"2\"\u003e\n \u003cp\u003e0.465\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\" rowspan=\"2\"\u003e\n \u003cp\u003e0.699\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\" rowspan=\"2\"\u003e\n \u003cp\u003e0.58\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\" rowspan=\"2\"\u003e\n \u003cp\u003e0.60\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e≤ 1.15 (mg/dL)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eref\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\u0026nbsp;\u003c/td\u003e\n \u003ctd align=\"left\" rowspan=\"2\"\u003e\n \u003cp\u003e\u0026lt; 0.001\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e\u0026gt; 1.15 (mg/dL)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e2.020\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e1.357–3.005\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\" rowspan=\"2\"\u003e\n \u003cp\u003ePaO\u003csub\u003e2\u003c/sub\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\" rowspan=\"2\"\u003e\n \u003cp\u003e0.612\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\" rowspan=\"2\"\u003e\n \u003cp\u003e0.559–0.666\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\" rowspan=\"2\"\u003e\n \u003cp\u003e\u0026lt; 0.001\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\" rowspan=\"2\"\u003e\n \u003cp\u003e68.45 (mmHg)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\" rowspan=\"2\"\u003e\n \u003cp\u003e0.645\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\" rowspan=\"2\"\u003e\n \u003cp\u003e0.571\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\" rowspan=\"2\"\u003e\n \u003cp\u003e0.64\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\" rowspan=\"2\"\u003e\n \u003cp\u003e0.57\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e\u0026gt; 68.45 (mmHg)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eref\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\u0026nbsp;\u003c/td\u003e\n \u003ctd align=\"left\" rowspan=\"2\"\u003e\n \u003cp\u003e\u0026lt; 0.001\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e≤ 68.45 (mmHg)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e2.416\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e1.634–3.574\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\" rowspan=\"2\"\u003e\n \u003cp\u003ePaO\u003csub\u003e2\u003c/sub\u003e/FiO\u003csub\u003e2\u003c/sub\u003e (OI)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\" rowspan=\"2\"\u003e\n \u003cp\u003e0.643\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\" rowspan=\"2\"\u003e\n \u003cp\u003e0.591–0.695\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\" rowspan=\"2\"\u003e\n \u003cp\u003e\u0026lt; 0.001\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\" rowspan=\"2\"\u003e\n \u003cp\u003e325.9524\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\" rowspan=\"2\"\u003e\n \u003cp\u003e0.735\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\" rowspan=\"2\"\u003e\n \u003cp\u003e0.509\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\" rowspan=\"2\"\u003e\n \u003cp\u003e0.56\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\" rowspan=\"2\"\u003e\n \u003cp\u003e0.71\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e\u0026gt; 325.9524\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eref\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\u0026nbsp;\u003c/td\u003e\n \u003ctd align=\"left\" rowspan=\"2\"\u003e\n \u003cp\u003e\u0026lt; 0.001\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e≤ 325.9524\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e2.416\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e1.634–3.574\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\" rowspan=\"2\"\u003e\n \u003cp\u003eSOFA\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\" rowspan=\"2\"\u003e\n \u003cp\u003e0.701\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\" rowspan=\"2\"\u003e\n \u003cp\u003e0.652–0.751\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\" rowspan=\"2\"\u003e\n \u003cp\u003e\u0026lt; 0.001\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\" rowspan=\"2\"\u003e\n \u003cp\u003e3\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\" rowspan=\"2\"\u003e\n \u003cp\u003e0.800\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\" rowspan=\"2\"\u003e\n \u003cp\u003e0.465\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\" rowspan=\"2\"\u003e\n \u003cp\u003e0.64\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\" rowspan=\"2\"\u003e\n \u003cp\u003e0.70\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e≤ 3\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eref\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\u0026nbsp;\u003c/td\u003e\n \u003ctd align=\"left\" rowspan=\"2\"\u003e\n \u003cp\u003e\u0026lt; 0.001\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e\u0026gt; 3\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e4.221\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e2.759–6.459\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\" rowspan=\"2\"\u003e\n \u003cp\u003eCO-RADS\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\" rowspan=\"2\"\u003e\n \u003cp\u003e0.608\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\" rowspan=\"2\"\u003e\n \u003cp\u003e0.555–0.661\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\" rowspan=\"2\"\u003e\n \u003cp\u003e\u0026lt; 0.001\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\" rowspan=\"2\"\u003e\n \u003cp\u003e4\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\" rowspan=\"2\"\u003e\n \u003cp\u003e0.807\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\" rowspan=\"2\"\u003e\n \u003cp\u003e0.410\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\" rowspan=\"2\"\u003e\n \u003cp\u003e0.57\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\" rowspan=\"2\"\u003e\n \u003cp\u003e0.70\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e≤ 4\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eref\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\u0026nbsp;\u003c/td\u003e\n \u003ctd align=\"left\" rowspan=\"2\"\u003e\n \u003cp\u003e\u0026lt; 0.001\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e\u0026gt; 4\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e3.399\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e2.125–5.439\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003c/tbody\u003e\n \u003c/table\u003e\n \u003c/div\u003e\n \u003cp\u003e*PPV: Positive predictive value, **NPV: Negative predictive value\u0026nbsp;\u003c/p\u003e\n \u003cdiv\u003e\n \u003ctable id=\"Tab3\" border=\"1\"\u003e\n \u003ccaption language=\"En\"\u003e\n \u003cdiv\u003eTable 4\u003c/div\u003e\n \u003cdiv\u003e\n \u003cp\u003eArea under the curve of multivariable prediction models\u003c/p\u003e\n \u003c/div\u003e\n \u003c/caption\u003e\n \u003cthead\u003e\n \u003ctr\u003e\n \u003cth align=\"left\"\u003e\n \u003cp\u003eModels\u003c/p\u003e\n \u003c/th\u003e\n \u003cth align=\"left\" colspan=\"5\"\u003e\n \u003cp\u003eAUC\u003c/p\u003e\n \u003c/th\u003e\n \u003c/tr\u003e\n \u003c/thead\u003e\n \u003ctbody\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\u0026nbsp;\u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eLR\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eGBDT\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eAdaBoost\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eXGBoost\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eCatBoost\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eIL-6 + LDH\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e0.801\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e0.896\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e0.874\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e0.887\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e0.892\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eAge + IL-6 + LDH\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e0.810\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e0.901\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e0.891\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e0.877\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e0.895\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eAge + IL-6 + SOFA\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e0.804\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e0.865\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e0.868\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e0.847\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e0.869\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eIL-6 + LDH + SOFA\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e0.830\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e0.907\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e0.897\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e0.887\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e0.902\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eAge + IL-6 + SOFA + CO-RADS\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e0.855\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e0.884\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e0.884\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e0.872\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e0.892\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eAge + IL-6 + LDH + SOFA\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e0.831\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e0.910\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e0.906\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e0.884\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e0.908\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eAge + IL-6 + LDH + SOFA + CO-RADS\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e0.873\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e0.928\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e0.919\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e0.914\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e0.925\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eAge + IL-6 + LDH + SOFA + CO-RADS + ESR\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e0.877\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e0.953\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e0.960\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e0.941\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e0.952\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003c/tbody\u003e\n \u003ctfoot\u003e\n \u003ctr\u003e\n \u003ctd colspan=\"6\"\u003eLR: Logistic regression, GBDT: gradient boosting decision tree, AdaBoost: adaptive boosting, XGBoost: eXtreme Gradient Boosting, CatBoost: categorical boosting, AUC: area under the receiver operating characteristic curve\u003c/td\u003e\n \u003c/tr\u003e\n \u003c/tfoot\u003e\n \u003c/table\u003e\n \u003c/div\u003e\n \u003cdiv\u003e\n \u003ctable id=\"Tab4\" border=\"1\"\u003e\n \u003ccaption language=\"En\"\u003e\n \u003cdiv\u003eTable 5\u003c/div\u003e\n \u003cdiv\u003e\n \u003cp\u003eParameters in the various model for (Age + IL-6 + LDH + SOFA + CO-RADS + ESR) prediction model\u003c/p\u003e\n \u003c/div\u003e\n \u003c/caption\u003e\n \u003cthead\u003e\n \u003ctr\u003e\n \u003cth align=\"left\"\u003e\u0026nbsp;\u003c/th\u003e\n \u003cth align=\"left\"\u003e\n \u003cp\u003eLR\u003c/p\u003e\n \u003c/th\u003e\n \u003cth align=\"left\"\u003e\n \u003cp\u003eGBDT\u003c/p\u003e\n \u003c/th\u003e\n \u003cth align=\"left\"\u003e\n \u003cp\u003eAdaBoost\u003c/p\u003e\n \u003c/th\u003e\n \u003cth align=\"left\"\u003e\n \u003cp\u003eXGBoost\u003c/p\u003e\n \u003c/th\u003e\n \u003cth align=\"left\"\u003e\n \u003cp\u003eCatBoost\u003c/p\u003e\n \u003c/th\u003e\n \u003c/tr\u003e\n \u003c/thead\u003e\n \u003ctbody\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eAUC\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e0.877\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e0.953\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e0.960\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e0.941\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e0.952\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e95% CI of AUC\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e0.840–0.908\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e0.933–0.970\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e0.941–0.974\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e0.919–0.961\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e0.932–0.969\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eBrier score\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e0.141\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e0.084\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e0.163\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e0.102\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e0.086\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eYouden index\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e0.594\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e0.596\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e0.480\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e0.624\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e0.523\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eSensitivity\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e0.690\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e0.870\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e0.915\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e0.875\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e0.875\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eSpecificity\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e0.898\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e0.916\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e0.885\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e0.889\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e0.903\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003ePPV\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e0.857\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e0.902\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e0.876\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e0.875\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e0.888\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eNPV\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e0.766\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e0.888\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e0.922\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e0.889\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e0.891\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003c/tbody\u003e\n \u003ctfoot\u003e\n \u003ctr\u003e\n \u003ctd colspan=\"6\"\u003eLR: Logistic regression, GBDT: gradient boosting decision tree, AdaBoost: adaptive boosting, XGBoost: eXtreme Gradient Boosting, CatBoost: categorical boosting, AUC: area under the receiver operating characteristic curve, CI: confidence interval, PPV: Positive predictive value, NPV: Negative predictive value\u003c/td\u003e\n \u003c/tr\u003e\n \u003c/tfoot\u003e\n \u003c/table\u003e\n \u003c/div\u003e\n \u003cp\u003eThe Kolmogorov-Smirnov test was used to assess the normality of quantitative variables. Descriptive statistics were presented using frequencies and percentages for qualitative variables, and means, standard deviations, medians, and interquartile ranges for quantitative variables. The independent t-test was used to compare the means of normally distributed quantitative variables between study groups. In contrast, the Mann-Whitney test was used to compare the median of non-normally distributed variables between groups. The chi-square or Fisher's exact test was used to compare qualitative variables between groups. The Pearson's correlation coefficient test was used to assess the correlation of quantitative variables in this study. The logistic regression test investigated the relationship between the studied variables and the disease outcome (mortality rate). The results of this test were expressed as a p-value, odds ratio (OR), and 95% confidence interval for OR. The Receiver Operating Characteristic (ROC) test was used to determine the cut-off point for determining mortality. The results of this statistical analysis were expressed using the area under the curve (AUC) and the 95% confidence interval for the AUC. Youden's index was also used to determine the cut-off point, sensitivity, and specificity. The Combined ROC curve was analyzed to integrate the predictability of multiple variables and introduced nine multi-variable models. We aimed to introduce the highest AUC and predictability, while involving the minimum number of variables. This has now been clarified in the revised manuscript. The multicollinearity among the predictors in the multivariable ROC model was assessed using the Variance Inflation Factor (VIF) in Python. All VIF values were below 1.1, indicating the absence of significant collinearity between the variables. A significance level of less than 0.05 was considered in this study.\u003c/p\u003e\n\u003c/div\u003e\n\u003ch3\u003eDevelopment and validation of the Models\u003c/h3\u003e\n\u003cp\u003eWe employed a conventional logistic regression method and popular machine learning classification algorithms, including adaptive boosting (AdaBoost), gradient boosting decision tree (GBDT), eXtreme Gradient Boosting (XGBoost), and categorical boosting (CatBoost) to develop predictive models and perform the internal validation via combining the 10-fold cross-validation and bootstrap resampling. The models were designed to involve two to six parameters, aiming to create the best AUC with the minimum number of parameters. The best model was determined according to the AUC comparison.\u003c/p\u003e\n\u003cp\u003eInternal validation was performed using the combination of 10-fold cross-validation and a bootstrap resampling approach to ensure robust estimation of model performance and mitigate potential overfitting. For each fold, models were trained on 90% of the random data and evaluated on the remaining 10%. Within each fold, we applied bootstrap resampling with 1,000 iterations to the test set to estimate confidence intervals for performance metrics, including AUC, sensitivity, and specificity. Predicted probabilities across all folds were then pooled to construct an overall ROC. Interpretation and evaluation of the risk prediction model were performed using calibration curves. The Youden index, sensitivity, and specificity at the cut-off point were determined. These analyses were performed using Python (version 3.9) with scikit-learn, XGBoost, and CatBoost packages.\u003c/p\u003e"},{"header":"Results","content":"\u003cdiv id=\"Sec7\" class=\"Section2\"\u003e\u003ch2\u003eStudy participants\u003c/h2\u003e\u003cp\u003eA total of 426 COVID-19 patients admitted to the ICU were studied. Of these, 226 (52.9%) patients were discharged, and 200 (47.1%) patients were deceased. The demographic and clinical features of the discharged and deceased groups are recorded in Table\u0026nbsp;\u003cspan refid=\"Tab1\" class=\"InternalRef\"\u003e1\u003c/span\u003e. The mean age was 61.51 years, which was significantly higher in the deceased group compared to the discharged group (65.96 vs. 57.57; P\u0026thinsp;\u0026lt;\u0026thinsp;0.001).\u003c/p\u003e\u003cp\u003eThe study population consisted of 224 (52.6%) male and 202 (47.4%) female patients. Male patients were more affected; however, there was no significant difference in gender between the deceased and discharged patients. RT-PCR testing was reported as positive for COVID-19 in 292 (68.54%) patients.\u003c/p\u003e\u003cp\u003eHypertension was the most common comorbidity (40.8%), followed by diabetes mellitus (22.53%), in the overall and the two groups. Among the 48 patients who required mechanical ventilation support, 33 were deceased, and 15 were discharged. There were significant differences between the discharged and deceased groups in terms of the prevalence of hypertension (p\u0026thinsp;=\u0026thinsp;0.001), diabetes mellitus comorbidities (p\u0026thinsp;=\u0026thinsp;0.001), and the number of patients who needed mechanical ventilation (p\u0026thinsp;=\u0026thinsp;0.001).\u003c/p\u003e\u003cp\u003eThe mean SOFA score on admission was 4.35, indicating a significant difference between the discharged and deceased groups (3.58 vs. 5.23; P\u0026thinsp;\u0026lt;\u0026thinsp;0.001). The CO-RADS score also showed a significant difference between the discharged and deceased groups (5.03 vs. 5.54; P\u0026thinsp;\u0026lt;\u0026thinsp;0.001).\u003c/p\u003e\u003cp\u003eIn Table\u0026nbsp;\u003cspan refid=\"Tab5\" class=\"InternalRef\"\u003e2\u003c/span\u003e, the mean and standard deviation values are recorded for the measured parameters. The odds ratio of parameters in association with mortality was calculated using logistic regression analysis, and the OR of the following parameters was higher than 1 (OR\u0026thinsp;\u0026gt;\u0026thinsp;1) and significant (P\u0026thinsp;\u0026lt;\u0026thinsp;0.05): 1. HR (OR: 1.017, P value\u0026thinsp;\u0026lt;\u0026thinsp;0.001), 2. RR (OR:1.053, P value\u0026thinsp;=\u0026thinsp;0.005), 3. IL-6 (OR: 1.023, P value\u0026thinsp;\u0026lt;\u0026thinsp;0.001), 4. ESR (OR:1.011, P value\u0026thinsp;\u0026lt;\u0026thinsp;0.001), 5. CRP (OR:1.011, P value\u0026thinsp;\u0026lt;\u0026thinsp;0.001), 6. LDH (OR: 1.010, P value\u0026thinsp;\u0026lt;\u0026thinsp;0.001), 7. Total bilirubin (OR: 1.319, P value: 0.020), 8. Creatinine (OR: 1.147, P value: 0.167), 9. SOFA score (OR: 1.345, P value\u0026thinsp;\u0026lt;\u0026thinsp;0.001), 10. CO-RADS (OR: 1.534, P value\u0026thinsp;\u0026lt;\u0026thinsp;0.001).\u003c/p\u003e\u003cp\u003e\u003cdiv class=\"gridtable\"\u003e\u003ctable float=\"Yes\" id=\"Tab5\" border=\"1\"\u003e\u003ccaption language=\"En\"\u003e\u003cdiv class=\"CaptionNumber\"\u003eTable 2\u003c/div\u003e\u003cdiv class=\"CaptionContent\"\u003e\u003cp\u003eClinical and laboratory measurements and their association with mortality\u003c/p\u003e\u003c/div\u003e\u003c/caption\u003e\u003ccolgroup cols=\"8\"\u003e\u003cdiv align=\"left\" class=\"colspec\" colname=\"c1\" colnum=\"1\"\u003e\u003c/div\u003e\u003cdiv align=\"left\" class=\"colspec\" colname=\"c2\" colnum=\"2\"\u003e\u003c/div\u003e\u003cdiv align=\"left\" class=\"colspec\" colname=\"c3\" colnum=\"3\"\u003e\u003c/div\u003e\u003cdiv align=\"left\" class=\"colspec\" colname=\"c4\" colnum=\"4\"\u003e\u003c/div\u003e\u003cdiv align=\"left\" class=\"colspec\" colname=\"c5\" colnum=\"5\"\u003e\u003c/div\u003e\u003cdiv align=\"left\" class=\"colspec\" colname=\"c6\" colnum=\"6\"\u003e\u003c/div\u003e\u003cdiv align=\"left\" class=\"colspec\" colname=\"c7\" colnum=\"7\"\u003e\u003c/div\u003e\u003cdiv align=\"char\" char=\".\" class=\"colspec\" colname=\"c8\" colnum=\"8\"\u003e\u003c/div\u003e\u003cthead\u003e\u003ctr\u003e\u003cth align=\"left\" colname=\"c1\"\u003e\u0026nbsp;\u003c/th\u003e\u003cth align=\"left\" colname=\"c2\"\u003e\u0026nbsp;\u003c/th\u003e\u003cth align=\"left\" colname=\"c3\"\u003e\u003cp\u003eDischarged\u003c/p\u003e\u003c/th\u003e\u003cth align=\"left\" colname=\"c4\"\u003e\u003cp\u003eDeceased\u003c/p\u003e\u003c/th\u003e\u003cth align=\"left\" colname=\"c5\"\u003e\u0026nbsp;\u003c/th\u003e\u003cth align=\"left\" colspan=\"3\" nameend=\"c8\" namest=\"c6\"\u003e\u003cp\u003eUnivariate logistic regression\u003c/p\u003e\u003c/th\u003e\u003c/tr\u003e\u003c/thead\u003e\u003ctbody\u003e\u003ctr\u003e\u003ctd align=\"left\" colspan=\"2\" nameend=\"c2\" namest=\"c1\"\u003e\u003cp\u003eVariables\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e\u003cp\u003eMean (SD*)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c4\"\u003e\u003cp\u003eMean (SD)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c5\"\u003e\u003cp\u003eP value\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c6\"\u003e\u003cp\u003eOR\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c7\"\u003e\u003cp\u003e95% CI\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c8\"\u003e\u003cp\u003eP value\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003eClinical indices\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u0026nbsp;\u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e\u0026nbsp;\u003c/td\u003e\u003ctd align=\"left\" colname=\"c4\"\u003e\u0026nbsp;\u003c/td\u003e\u003ctd align=\"left\" colname=\"c5\"\u003e\u0026nbsp;\u003c/td\u003e\u003ctd align=\"left\" colname=\"c6\"\u003e\u0026nbsp;\u003c/td\u003e\u003ctd align=\"left\" colname=\"c7\"\u003e\u0026nbsp;\u003c/td\u003e\u003ctd align=\"left\" colname=\"c8\"\u003e\u0026nbsp;\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u0026nbsp;\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003eHear rate\u003csup\u003e1\u003c/sup\u003e\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e\u003cp\u003e85.71 (17.369)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c4\"\u003e\u003cp\u003e91.45 (20.031)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c5\"\u003e\u003cp\u003e0.002*\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c6\"\u003e\u003cp\u003e1.017\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c7\"\u003e\u003cp\u003e1.006\u0026ndash;1.027\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c8\"\u003e\u003cp\u003e\u0026lt;\u0026thinsp;0.001*\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u0026nbsp;\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003eTemperature\u003csup\u003e2\u003c/sup\u003e (\u0026deg;C)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e\u003cp\u003e36.808 (0.540)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c4\"\u003e\u003cp\u003e36.762 (0.536)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c5\"\u003e\u003cp\u003e0.259\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c6\"\u003e\u003cp\u003e0.851\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c7\"\u003e\u003cp\u003e0.595\u0026ndash;1.217\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c8\"\u003e\u003cp\u003e\u0026lt;\u0026thinsp;0.001*\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u0026nbsp;\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003eRespiratory Rate**\u003csup\u003e2\u003c/sup\u003e (210/167)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e\u003cp\u003e21.13 (5.421)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c4\"\u003e\u003cp\u003e22.86 (6.208)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c5\"\u003e\u003cp\u003e0.002*\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c6\"\u003e\u003cp\u003e1.053\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c7\"\u003e\u003cp\u003e1.016\u0026ndash;1.091\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c8\"\u003e\u003cp\u003e0.005*\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u0026nbsp;\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003eSBP\u003csup\u003e1\u003c/sup\u003e (mmHg)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e\u003cp\u003e127.59 (21.888)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c4\"\u003e\u003cp\u003e127.59 (25.512)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c5\"\u003e\u003cp\u003e0.997\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c6\"\u003e\u003cp\u003e1.000\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c7\"\u003e\u003cp\u003e0.992\u0026ndash;1.008\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c8\"\u003e\u003cp\u003e0.997\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u0026nbsp;\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003eDBP\u003csup\u003e1\u003c/sup\u003e (mmHg)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e\u003cp\u003e78.06 (14.452)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c4\"\u003e\u003cp\u003e77.50 (17.296)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c5\"\u003e\u003cp\u003e0.716\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c6\"\u003e\u003cp\u003e0.998\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c7\"\u003e\u003cp\u003e0.086\u0026ndash;1.010\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c8\"\u003e\u003cp\u003e0.712\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u0026nbsp;\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003eMAP\u003csup\u003e1\u003c/sup\u003e (mmHg)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e\u003cp\u003e94.572 (15.666)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c4\"\u003e\u003cp\u003e94.191 (18.886)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c5\"\u003e\u003cp\u003e0.822\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c6\"\u003e\u003cp\u003e0.999\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c7\"\u003e\u003cp\u003e0.088\u0026ndash;1.010\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c8\"\u003e\u003cp\u003e0.820\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u0026nbsp;\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003eGCS\u003csup\u003e2\u003c/sup\u003e\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e\u003cp\u003e14.47 (1.424)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c4\"\u003e\u003cp\u003e13.28 (2.841)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c5\"\u003e\u003cp\u003e\u0026lt;\u0026thinsp;0.001*\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c6\"\u003e\u003cp\u003e0.768\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c7\"\u003e\u003cp\u003e0.692\u0026ndash;0.853\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c8\"\u003e\u003cp\u003e\u0026lt;\u0026thinsp;0.001*\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003eOxygen indices\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u0026nbsp;\u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e\u0026nbsp;\u003c/td\u003e\u003ctd align=\"left\" colname=\"c4\"\u003e\u0026nbsp;\u003c/td\u003e\u003ctd align=\"left\" colname=\"c5\"\u003e\u0026nbsp;\u003c/td\u003e\u003ctd align=\"left\" colname=\"c6\"\u003e\u0026nbsp;\u003c/td\u003e\u003ctd align=\"left\" colname=\"c7\"\u003e\u0026nbsp;\u003c/td\u003e\u003ctd align=\"left\" colname=\"c8\"\u003e\u0026nbsp;\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u0026nbsp;\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003ePaO\u003csub\u003e2\u003c/sub\u003e\u003csup\u003e2\u003c/sup\u003e (mmHg)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e\u003cp\u003e73.732 (22.891)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c4\"\u003e\u003cp\u003e66.297 (15.605)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c5\"\u003e\u003cp\u003e\u0026lt;\u0026thinsp;0.001*\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c6\"\u003e\u003cp\u003e0.978\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c7\"\u003e\u003cp\u003e0.966\u0026ndash;0.990\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c8\"\u003e\u003cp\u003e\u0026lt;\u0026thinsp;0.001*\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u0026nbsp;\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003eFiO\u003csub\u003e2\u003c/sub\u003e\u003csup\u003e2\u003c/sup\u003e (%)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e\u003cp\u003e26.59 (20.307)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c4\"\u003e\u003cp\u003e34.03 (29.397)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c5\"\u003e\u003cp\u003e0.003*\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c6\"\u003e\u003cp\u003e1.012\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c7\"\u003e\u003cp\u003e1.004\u0026ndash;1.020\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c8\"\u003e\u003cp\u003e0.003*\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u0026nbsp;\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003ePaO\u003csub\u003e2\u003c/sub\u003e/FiO\u003csub\u003e2\u003c/sub\u003e (OI)\u003csup\u003e2\u003c/sup\u003e\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e\u003cp\u003e321.147 (94.186)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c4\"\u003e\u003cp\u003e269.308 (103.142)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c5\"\u003e\u003cp\u003e\u0026lt;\u0026thinsp;0.001*\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c6\"\u003e\u003cp\u003e0.995\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c7\"\u003e\u003cp\u003e0.992\u0026ndash;0.997\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c8\"\u003e\u003cp\u003e\u0026lt;\u0026thinsp;0.001*\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003eLaboratory indices\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u0026nbsp;\u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e\u0026nbsp;\u003c/td\u003e\u003ctd align=\"left\" colname=\"c4\"\u003e\u0026nbsp;\u003c/td\u003e\u003ctd align=\"left\" colname=\"c5\"\u003e\u0026nbsp;\u003c/td\u003e\u003ctd align=\"left\" colname=\"c6\"\u003e\u0026nbsp;\u003c/td\u003e\u003ctd align=\"left\" colname=\"c7\"\u003e\u0026nbsp;\u003c/td\u003e\u003ctd align=\"left\" colname=\"c8\"\u003e\u0026nbsp;\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u0026nbsp;\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003eWBC (\u0026times; 10\u003csup\u003e3\u003c/sup\u003e cells/\u0026micro;L)\u003csup\u003e2\u003c/sup\u003e\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e\u003cp\u003e11.257 (17.484)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c4\"\u003e\u003cp\u003e11.739 (8.432)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c5\"\u003e\u003cp\u003e0.049*\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c6\"\u003e\u003cp\u003e1.003\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c7\"\u003e\u003cp\u003e0.989\u0026ndash;1.017\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c8\"\u003e\u003cp\u003e0.725\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u0026nbsp;\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003eLymphocyte cells (\u0026times; 10\u003csup\u003e3\u003c/sup\u003e cells/\u0026micro;L)\u003csup\u003e2\u003c/sup\u003e\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e\u003cp\u003e1.556 (9.283)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c4\"\u003e\u003cp\u003e1.529 (5.078)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c5\"\u003e\u003cp\u003e0.434\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c6\"\u003e\u003cp\u003e0.999\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c7\"\u003e\u003cp\u003e0.975\u0026ndash;1.025\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c8\"\u003e\u003cp\u003e0.961\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u0026nbsp;\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003eLymphocyte (%)\u003csup\u003e2\u003c/sup\u003e\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e\u003cp\u003e9.413 (7.157)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c4\"\u003e\u003cp\u003e10.191 (9.532)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c5\"\u003e\u003cp\u003e0.641\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c6\"\u003e\u003cp\u003e1.011\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c7\"\u003e\u003cp\u003e0.988\u0026ndash;1.035\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c8\"\u003e\u003cp\u003e0.340\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u0026nbsp;\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003ePlatelet cells (\u0026times; 10\u003csup\u003e3\u003c/sup\u003e cells/\u0026micro;L)\u003csup\u003e2\u003c/sup\u003e\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e\u003cp\u003e225.94 (101.130)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c4\"\u003e\u003cp\u003e210.82 (95.561)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c5\"\u003e\u003cp\u003e0.087\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c6\"\u003e\u003cp\u003e0.998\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c7\"\u003e\u003cp\u003e0.996-1.000\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c8\"\u003e\u003cp\u003e0.116\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u0026nbsp;\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003eIL-6\u003csup\u003e2\u003c/sup\u003e (pg/mL)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e\u003cp\u003e15.656 (67.214)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c4\"\u003e\u003cp\u003e52.822 (58.862)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c5\"\u003e\u003cp\u003e\u0026lt;\u0026thinsp;0.001*\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c6\"\u003e\u003cp\u003e1.023\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c7\"\u003e\u003cp\u003e1.016\u0026ndash;1.031\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c8\"\u003e\u003cp\u003e\u0026lt;\u0026thinsp;0.001*\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u0026nbsp;\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003eESR\u003csup\u003e2\u003c/sup\u003e (mm/hr)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e\u003cp\u003e56.527 (21.289)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c4\"\u003e\u003cp\u003e70.979 (48.008)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c5\"\u003e\u003cp\u003e0.118\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c6\"\u003e\u003cp\u003e1.011\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c7\"\u003e\u003cp\u003e1.006\u0026ndash;1.017\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c8\"\u003e\u003cp\u003e\u0026lt;\u0026thinsp;0.001*\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u0026nbsp;\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003eCRP \u003csup\u003e2\u003c/sup\u003e *** (mg/L)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e\u003cp\u003e48.694 (33.061)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c4\"\u003e\u003cp\u003e70.659 (61.235)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c5\"\u003e\u003cp\u003e0.002*\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c6\"\u003e\u003cp\u003e1.011\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c7\"\u003e\u003cp\u003e1.006\u0026ndash;1.016\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c8\"\u003e\u003cp\u003e\u0026lt;\u0026thinsp;0.001*\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u0026nbsp;\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003eLDH\u003csup\u003e2\u003c/sup\u003e (U/L)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e\u003cp\u003e393.85 (83.974)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c4\"\u003e\u003cp\u003e482.85 (103.064)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c5\"\u003e\u003cp\u003e\u0026lt;\u0026thinsp;0.001*\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c6\"\u003e\u003cp\u003e1.010\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c7\"\u003e\u003cp\u003e1.007\u0026ndash;1.012\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c8\"\u003e\u003cp\u003e\u0026lt;\u0026thinsp;0.001*\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u0026nbsp;\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003eBilirubin (Total)\u003csup\u003e2\u003c/sup\u003e (mg/dL)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e\u003cp\u003e1.134 (0.879)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c4\"\u003e\u003cp\u003e1.356 (0.995)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c5\"\u003e\u003cp\u003e0.003*\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c6\"\u003e\u003cp\u003e1.319\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c7\"\u003e\u003cp\u003e1.044\u0026ndash;1.666\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c8\"\u003e\u003cp\u003e0.020*\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u0026nbsp;\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003eCreatinine\u003csup\u003e2\u003c/sup\u003e (mg/dL)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e\u003cp\u003e1.208 (0.938)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c4\"\u003e\u003cp\u003e1.353 (1.155)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c5\"\u003e\u003cp\u003e0.090\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c6\"\u003e\u003cp\u003e1.147\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c7\"\u003e\u003cp\u003e0.944\u0026ndash;1.393\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c8\"\u003e\u003cp\u003e0.167\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003eScoring systems\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003eSOFA\u003csup\u003e2\u003c/sup\u003e\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e\u003cp\u003e3.58 (2.034)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c4\"\u003e\u003cp\u003e5.23 (2.766)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c5\"\u003e\u003cp\u003e\u0026lt;\u0026thinsp;0.001*\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c6\"\u003e\u003cp\u003e1.345\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c7\"\u003e\u003cp\u003e1.226\u0026ndash;1.477\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c8\"\u003e\u003cp\u003e\u0026lt;\u0026thinsp;0.001*\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u0026nbsp;\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003eCO-RADS\u003csup\u003e2\u003c/sup\u003e\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e\u003cp\u003e5.03 (1.194)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c4\"\u003e\u003cp\u003e5.54 (0.961)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c5\"\u003e\u003cp\u003e\u0026lt;\u0026thinsp;0.001*\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c6\"\u003e\u003cp\u003e1.534\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c7\"\u003e\u003cp\u003e1.275\u0026ndash;1.844\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c8\"\u003e\u003cp\u003e\u0026lt;\u0026thinsp;0.001*\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003c/tbody\u003e\u003c/colgroup\u003e\u003ctfoot\u003e\u003ctr\u003e\u003ctd colspan=\"8\"\u003e*Significant differences (p‑value\u0026thinsp;\u0026lt;\u0026thinsp;0.05), **Respiratory rate data were available for 210 discharged and 167 deceased patients, ***CRP data were available for 226 discharged and 144 deceased patients. Other parameters were recorded for all 226 discharged and 200 deceased patients. SD Standard Deviation, OR Odds Ratio, CI Confidence interval, SBP systolic blood pressure, DBP diastolic blood pressure, MAP mean arterial pressure, GCS Glasgow Coma Scale, PaO\u003csub\u003e2\u003c/sub\u003e arterial partial pressure of oxygen, FiO\u003csub\u003e2\u003c/sub\u003e fraction of inspiratory oxygen concentration, OI oxygenation index, WBC white blood cell count, IL-6 interleukin 6, ESR erythrocytes sedimentation rate, CRP c-reactive protein, LDH lactate dehydrogenase, SOFA sequential organ failure assessment, CO-RADS COVID-19 Reporting and Data System, \u003csup\u003e1/\u003c/sup\u003e T-test, \u003csup\u003e2/\u003c/sup\u003e Mann\u0026ndash;Whitney U test.\u003c/td\u003e\u003c/tr\u003e\u003c/tfoot\u003e\u003c/table\u003e\u003c/div\u003e\u003c/p\u003e\u003cp\u003eFindings indicate that a one-point increase in SOFA and CO-RADS scores corresponded to a 34.5% and 53.4% increase in mortality risk, respectively. A one-point rise in GCS was associated with a 23.2% decrease in mortality risk. In addition to the clinical findings, the odds ratios of inflammatory markers are listed in Table\u0026nbsp;\u003cspan refid=\"Tab5\" class=\"InternalRef\"\u003e2\u003c/span\u003e.\u003c/p\u003e\u003cp\u003eAccording to Table\u0026nbsp;\u003cspan refid=\"Tab2\" class=\"InternalRef\"\u003e3\u003c/span\u003e and ROC curves in Fig.\u0026nbsp;\u003cspan refid=\"Fig1\" class=\"InternalRef\"\u003e1\u003c/span\u003e, the cut-off points for mortality were measured for the following parameters using the Youden index: Age (Fig.\u0026nbsp;\u003cspan refid=\"Fig1\" class=\"InternalRef\"\u003e1\u003c/span\u003eA) (\u0026gt;\u0026thinsp;50 years, AUC:0.633, 87% sensitivity, 34.7% specificity), CRP (Fig.\u0026nbsp;\u003cspan refid=\"Fig1\" class=\"InternalRef\"\u003e1\u003c/span\u003eB) (\u0026gt;\u0026thinsp;70 mg/L, AUC:0.597, 38.9% sensitivity, 81.4% specificity), IL-6 (Fig.\u0026nbsp;\u003cspan refid=\"Fig1\" class=\"InternalRef\"\u003e1\u003c/span\u003eC) (\u0026gt;\u0026thinsp;18.95 pg/mL, AUC: 0.761, 65.5% sensitivity, 93.4% specificity), LDH (Fig.\u0026nbsp;\u003cspan refid=\"Fig1\" class=\"InternalRef\"\u003e1\u003c/span\u003eD) (\u0026gt;\u0026thinsp;437.5 U/L, AUC: 0.737, 63% sensitivity, 68.6% specificity), WBC (Fig.\u0026nbsp;\u003cspan refid=\"Fig1\" class=\"InternalRef\"\u003e1\u003c/span\u003eE) (\u0026gt;\u0026thinsp;10.15\u0026times;10\u003csup\u003e9\u003c/sup\u003e/L, AUC: 0.555, 54% sensitivity, 57.1% specificity), Bilirubin (Total) (Fig.\u0026nbsp;\u003cspan refid=\"Fig1\" class=\"InternalRef\"\u003e1\u003c/span\u003eF) (\u0026gt;\u0026thinsp;1.15 mg/dL, AUC: 0.582, 46.5% sensitivity, 69.9% specificity), PaO\u003csub\u003e2\u003c/sub\u003e (Fig.\u0026nbsp;\u003cspan refid=\"Fig1\" class=\"InternalRef\"\u003e1\u003c/span\u003eG) (\u0026le;\u0026thinsp;68.45 mmHg, AUC: 0.612, 64.5% sensitivity, 57.1% specificity), PaO\u003csub\u003e2\u003c/sub\u003e/FiO\u003csub\u003e2\u003c/sub\u003e (Oxygenation index: OI) (Fig.\u0026nbsp;\u003cspan refid=\"Fig1\" class=\"InternalRef\"\u003e1\u003c/span\u003eH) (\u0026le;\u0026thinsp;325.95, AUC: 0.643, 50.9% sensitivity, 73.5% specificity), SOFA (Fig.\u0026nbsp;\u003cspan refid=\"Fig1\" class=\"InternalRef\"\u003e1\u003c/span\u003eI) (\u0026gt;\u0026thinsp;3, AUC: 0.701, 80% sensitivity, 46.5% specificity) (Fig.\u0026nbsp;\u003cspan refid=\"Fig1\" class=\"InternalRef\"\u003e1\u003c/span\u003eJ) (\u0026gt;\u0026thinsp;18.5, AUC: 0.662, 66% sensitivity, 55.8% specificity), CO-RADS (Fig.\u0026nbsp;\u003cspan refid=\"Fig1\" class=\"InternalRef\"\u003e1\u003c/span\u003eK) (\u0026gt;\u0026thinsp;4, AUC:0.608, 80.7% sensitivity, 41% specificity).\u003c/p\u003e\u003c/div\u003e\u003cdiv id=\"Sec8\" class=\"Section2\"\u003e\u003ch2\u003eMachine learning models performance and validation\u003c/h2\u003e\u003cp\u003eThe AUC of multi-factor models was measured using logistic regression and four ML-based algorithms. Models with an AUC\u0026thinsp;\u0026gt;\u0026thinsp;0.800 were recorded in Table\u0026nbsp;\u003cspan refid=\"Tab3\" class=\"InternalRef\"\u003e4\u003c/span\u003e, and related ROCs are demonstrated in Fig.\u0026nbsp;\u003cspan refid=\"Fig2\" class=\"InternalRef\"\u003e2\u003c/span\u003e. According to Table\u0026nbsp;\u003cspan refid=\"Tab3\" class=\"InternalRef\"\u003e4\u003c/span\u003e, the six-factor model, which included age, IL-6, LDH, SOFA, CO-RADS, and ESR, presented the highest AUC in each setting compared to the other models. In terms of prediction performance, the results of the logistic regression and four machine learning algorithms on the test set indicated AUCs of 0.960 for AdaBoost, 0.953 for GBDT, 0.952 for CatBoost, 0.941 for XGBoost, and 0.877 for logistic regression. The AUC, 95% CI, Brier score, Youden index, sensitivity, specificity, PPV, and NPV values of each model in the test data set are listed in Table\u0026nbsp;\u003cspan refid=\"Tab4\" class=\"InternalRef\"\u003e5\u003c/span\u003e. The calibration curves for internal validation for each model are shown in Fig.\u0026nbsp;\u003cspan refid=\"Fig3\" class=\"InternalRef\"\u003e3\u003c/span\u003e. The AdaBoost model demonstrated the best performance with an AUC of 0.960 (95% CI: 0.941\u0026ndash;0.974). The Youden index was 0.480, and sensitivity and specificity at the optimum cut-off point were reported as 0.915 and 0.885, respectively. According to combined internal validation, the GBDT algorithm demonstrated the best calibration, as indicated by the Brier score (0.084).\u003c/p\u003e\u003c/div\u003e"},{"header":"Discussion","content":"\u003cp\u003eWe included 426 ICU-admitted COVID-19 patients and investigated their mortality, clinical features, and biomarkers in two groups: the deceased and discharged groups. The male-to-female ratio was 1.1:1 (224 males vs. 202 females), and the mortality rate was 47.1% (n\u0026thinsp;=\u0026thinsp;200).\u003c/p\u003e\u003cp\u003eHypertension and diabetes mellitus were the most common comorbidities in both groups, and there was a significant correlation between these comorbidities and the mortality rate. In a systematic review and meta-analysis of the findings of 10 articles (76,993 patients), the prevalence of hypertension, cardiovascular diseases, cancer, chronic kidney failure, smoking, and diabetes was significantly higher in COVID-19 patients compared to non-affected patients (\u003cspan citationid=\"CR19\" class=\"CitationRef\"\u003e19\u003c/span\u003e). In a systematic review of the findings of 114 studies (310,494 patients), among the 72 predictive variables, advanced age, hypertension, and diabetes were reported as the most important predictors of mortality (based on the 10% cut-off point). Diabetes was identified as an independent risk factor for mortality in COVID-19 patients (\u003cspan citationid=\"CR20\" class=\"CitationRef\"\u003e20\u003c/span\u003e). In a community study of 3,862,012 COVID-19 patients in England, cardiovascular diseases, diabetes, a history of steroid treatment, obesity, chronic kidney disease, chronic obstructive pulmonary disease, and immunodeficiency diseases were significantly associated with mortality from COVID-19 in patients aged 70 years or older (\u003cspan citationid=\"CR21\" class=\"CitationRef\"\u003e21\u003c/span\u003e).\u003c/p\u003e\u003cp\u003eIn our study, deceased patients were significantly older than the discharged group. The mortality risk for patients older than 50 was 3.366 times higher than that for younger patients. In agreement with our findings, a systematic review study encompassing 49 studies, which included more than 587,000 COVID-19 patients, reported that advanced age (defined as being older than 50 years) and comorbidities significantly increased the risk of in-hospital mortality in COVID-19 patients. Of note, the risk of hospital mortality in patients over 60 years old was 23 times higher than that of patients who were less than 60 years old (\u003cspan citationid=\"CR22\" class=\"CitationRef\"\u003e22\u003c/span\u003e). In another systematic review and meta-analysis of the data from five countries, China, Italy, Spain, England, and the USA (New York state) (611,583 patients in total), a significant association was reported between age and mortality rate and nosocomial mortality in patients with COVID-19 (\u003cspan citationid=\"CR23\" class=\"CitationRef\"\u003e23\u003c/span\u003e). The mortality rate in ICU-admitted COVID-19 patients is variable depending on the region and health system. It has been reported that higher altitudes are associated with enhanced survival rates of COVID-19 patients (\u003cspan citationid=\"CR24\" class=\"CitationRef\"\u003e24\u003c/span\u003e) and decreased COVID-19 fatality (\u003cspan citationid=\"CR25\" class=\"CitationRef\"\u003e25\u003c/span\u003e). This study reported a 47.1% mortality rate in ICU-admitted COVID-19 patients in Zanjan, Iran, where the altitude is 1638 meters above sea level. Similar studies on the Iranian population in an identical health system, conducted on 204 and 133 ICU-admitted COVID-19 patients in Tehran, Iran (at a lower elevation of 1200 m), indicated higher mortality rates (55.9% and 57.9%, respectively) (\u003cspan citationid=\"CR26\" class=\"CitationRef\"\u003e26\u003c/span\u003e, \u003cspan citationid=\"CR27\" class=\"CitationRef\"\u003e27\u003c/span\u003e).\u003c/p\u003e\u003cp\u003eIn a study of 184,875 ICU COVID-19 patients, the SOFA score outperformed the qSOFA score and SIRS criteria, with AUCs of 0.753, 0.607, and 0.589, respectively (\u003cspan citationid=\"CR28\" class=\"CitationRef\"\u003e28\u003c/span\u003e). A study involving 204 ICU-admitted COVID-19 patients examined the use of SOFA and APACHE II scores in predicting mortality. The area under the curve was 89.5% for SOFA and 73% for the APACHE II score, and SOFA showed a higher predictive value compared to the APACHE II score (\u003cspan citationid=\"CR27\" class=\"CitationRef\"\u003e27\u003c/span\u003e). Hence, we investigated the predictability of the SOFA system in ICU-admitted patients with COVID-19. The current study also demonstrated that deceased patients had significantly higher CO-RADS and SOFA scores. In another survey of 134 COVID-19 patients, the CO-RADS scores and SOFA scores in ICU patients were considerably higher than those in patients admitted to clinical wards (\u003cspan citationid=\"CR29\" class=\"CitationRef\"\u003e29\u003c/span\u003e).\u003c/p\u003e\u003cp\u003eWe investigated the biomarkers and clinical features of COVID-19 patients and their potential in predicting mortality using ROC analyses and odds ratios. All clinical records were recorded on ICU admission. AUC-ROC was the highest for LDH (\u0026gt;\u0026thinsp;437.5 U/L, AUC: 0.737, 95% CI: 0.690\u0026ndash;0.783, OR: 3.717), followed by IL-6 (\u0026gt;\u0026thinsp;18.95 pg/mL, AUC: 0.761, 95% CI: 0.711\u0026ndash;0.810, OR: 26.706) and SOFA score (\u0026gt;\u0026thinsp;3, AUC: 0.701, CI:0.652\u0026ndash;0.751, OR: 4.221).\u003c/p\u003e\u003cp\u003eA study involving 1,003 COVID-19 patients in France aimed to combine deep learning with CT scans, biological, and clinical data to predict the severity of COVID-19 patients. The findings indicated that AI-driven CT scans outperformed those evaluated by radiologists for several reasons. First, AI-driven CT scans had a higher AUC of 0.75 compared to 0.66 from radiologists' evaluations for predicting disease severity in a 135-patient validation cohort from Institut Gustave Roussy (IGR). Additionally, AI can integrate clinical features such as age and sex and capture some clinical information when data is missing in its assessments. Radiologists must read numerous cases and prioritize COVID-19 patients, while AI can help alleviate this burden by processing data more efficiently (\u003cspan citationid=\"CR30\" class=\"CitationRef\"\u003e30\u003c/span\u003e).\u003c/p\u003e\u003cp\u003eRecent advancements in clinical practice, the emergence of new interventions and treatments, and the evolution of patient monitoring methods over the past 25 years have created a significant need to update the SOFA score (\u003cspan citationid=\"CR31\" class=\"CitationRef\"\u003e31\u003c/span\u003e).\u003c/p\u003e\u003cp\u003eIn a retrospective observational study involving 117 hospitalized patients with COVID-19, researchers sought to investigate the prognostic value of the SOFA score in these patients. The AUC for the diagnostic accuracy of the SOFA in predicting severe COVID-19 was 0.908, with a cutoff value of 2, a sensitivity of 85.20%, and a specificity of 80.40%. The SOFA score was assessed at the onset of the disease or upon admission (\u003cspan citationid=\"CR32\" class=\"CitationRef\"\u003e32\u003c/span\u003e). Another study of 320 patients with COVID-19 pneumonia who experienced severe respiratory distress employed ROC analysis to predict mortality, reporting an AUC of 0.883, a sensitivity of 74%, and a specificity of 80% at a cutoff point (SOFA scores\u0026thinsp;=\u0026thinsp;5). They assessed the worst SOFA score recorded within 48 hours of the onset of severe respiratory distress (\u003cspan citationid=\"CR16\" class=\"CitationRef\"\u003e16\u003c/span\u003e). In another study on SOFA in various infectious states, the authors evaluated the worst SOFA score within 24 hours of ICU admission (\u003cspan citationid=\"CR33\" class=\"CitationRef\"\u003e33\u003c/span\u003e). The calculation time for the SOFA score varies among other studies. In our study, the retrospective design and the lack of sequential SOFA score recording precluded the possibility of determining the worst SOFA score over time; hence, we assessed the SOFA score recorded upon ICU admission. In addition to the differences in study populations and designs among studies, the lack of worst SOFA records may explain the differences in determining the optimum cut-off point for the SOFA score.\u003c/p\u003e\u003cp\u003eIL-6 is recognized as a prototype cytokine that plays a significant role in host defense through its pleiotropic activities, which are released by various cell types. IL-6 stimulates immune responses and promotes acute-phase reactions as well as hematopoiesis (\u003cspan citationid=\"CR34\" class=\"CitationRef\"\u003e34\u003c/span\u003e). In our study, consistent with a previous study (\u003cspan citationid=\"CR35\" class=\"CitationRef\"\u003e35\u003c/span\u003e), we identified IL-6 as a significant predictor of mortality in COVID-19 patients, with an AUC of 0.761, CI: 0.711\u0026ndash;0.811, specificity of 98.4%, and sensitivity of 65.5% at a cut-off value of 18.95. A study by Liu et al. indicated that monitoring IL-6 levels alone can reliably predict disease severity, although it was not linked to mortality (\u003cspan citationid=\"CR36\" class=\"CitationRef\"\u003e36\u003c/span\u003e). Additionally, another study showed that IL-6 is a more effective predictor of the need for mechanical ventilation than mortality (\u003cspan citationid=\"CR37\" class=\"CitationRef\"\u003e37\u003c/span\u003e).\u003c/p\u003e\u003cp\u003eLDH is an intracellular enzyme that facilitates anaerobic glycolysis by converting pyruvate into lactate, which is routinely tested as a key indicator of inflammation (\u003cspan citationid=\"CR38\" class=\"CitationRef\"\u003e38\u003c/span\u003e). In a study on 203 COVID-19 patients, serum LDH level higher than 359.50 U/L predicted COVID-19 mortality with 93.8% sensitivity and 88.2% specificity, and LDH was identified as an independent predictor of COVID-19 mortality (\u003cspan citationid=\"CR39\" class=\"CitationRef\"\u003e39\u003c/span\u003e). In our study, which included ICU-admitted COVID-19 patients, the serum LDH level at the cut-off point (437.5 U/L) demonstrated 63.0% sensitivity and 68.6% specificity for predicting mortality.\u003c/p\u003e\u003cp\u003ePan et al. developed a machine learning-based model to predict mortality using eight factors: lymphocyte percentage, prothrombin time, LDH, total bilirubin, and eosinophil percentage, creatinine, neutrophil percentage, and albumin level. This model predicted 5-fold cross validation (AUC\u0026thinsp;=\u0026thinsp;0.86) and the verification queue (AUC\u0026thinsp;=\u0026thinsp;0.92). The best recognition performance was observed with an AUC of 0.86 in the 5-fold cross-validation training set, further improving to 0.92 in the validation cohort (\u003cspan citationid=\"CR40\" class=\"CitationRef\"\u003e40\u003c/span\u003e). Another prediction model incorporated key predictors, including age, chronic lung disease, CRP, D-dimer, NLR, creatinine, and total bilirubin. It showed intense discrimination in both the training (AUC\u0026thinsp;=\u0026thinsp;0.912, 95% CI: 0.884\u0026ndash;0.940) and validation cohorts (AUC\u0026thinsp;=\u0026thinsp;0.922, 95% CI: 0.891\u0026ndash;0.953), with good calibration (\u003cspan citationid=\"CR41\" class=\"CitationRef\"\u003e41\u003c/span\u003e). Previous studies have investigated the prognostic value of clinical parameters; however, there is limited research on integrating the predictive value of clinical, laboratory, and imaging findings to enhance COVID-19 mortality prediction in the intensive care setting. In the present study, models were designed incorporating clinical predictors and using ML algorithms. Eight models demonstrated an AUC higher than 0.800 in logistic regression and ML algorithms. Internal validation was performed using an integrated 10-fold stratified cross-validation with bootstrap. The six-factor model showed the best performance in the AdaBoost algorithm, with an AUC of 0.960 (95% CI: 0.941\u0026ndash;0.974). The combination of cross-validation with bootstrapped confidence intervals ensured robust internal validation and minimized the risk of overestimating model performance. Incorporating both discrimination and calibration metrics, including AUC and Brier score, offered a more comprehensive evaluation of clinical predictive models. Despite internal validation results, our models lack external validation. Additionally, the predictive model should be interpreted cautiously due to the limitations of retrospective data, and further validation in larger prospective cohorts is necessary before clinical implementation. Validating these models in independent cohorts is essential to confirm their performance across diverse clinical settings and to support their broader clinical application.\u003c/p\u003e"},{"header":"Limitations","content":"\u003cp\u003eThe retrospective nature of this study and the absence of a longitudinal cohort design imposed significant limitations on developing a prediction model. This may introduce potential selection and information biases that could affect the reliability of the results. Additionally, our prediction model was developed and tested in a single-center study, which raises the possibility of overfitting and limits its generalizability. Therefore, future prospective multicenter studies with external validation are needed to confirm and improve the predictive performance of our model.\u003c/p\u003e\u003cp\u003eDue to the retrospective design, not all variables were collected for every patient, and missing data were observed in some variables. Moreover, the lack of a longitudinal cohort design precluded the possibility of follow-up on patients and evaluating temporal changes in clinical parameters to assess their dynamic prognostic value over time. The lack of longitudinal data limited the ability to infer the causality between the predictors and mortality outcomes.\u003c/p\u003e\u003cp\u003eA prediction model development requires external validation (using independent datasets) and temporal validation to assess model performance over time. This study did not include external validation due to limited access to external datasets. As a result, our findings may not fully generalize to other patient populations or clinical settings.\u003c/p\u003e\u003cp\u003eThe incomplete patient records and the need for all the measurements required to calculate SOFA or CT scan scores. On the other hand, the limited prevalence of some underlying diseases prevented researchers from conducting more accurate statistical tests; therefore, it was impossible to investigate the effect of these underlying diseases on disease outcomes. It is recommended that further studies examine the impact of these diseases on COVID-19 outcomes as well as imaging findings, using a case-control design.\u003c/p\u003e"},{"header":"Conclusion","content":"\u003cp\u003eClinical parameters, including IL-6, LDH, and SOFA, showed a good predictive value for mortality in ICU-admitted COVID-19 patients. Multi-organ assessment using the SOFA score showed a greater AUC for mortality predictability compared to the CT-based CO-RADS system in critical care. Machine learning models have enhanced the predictability of mortality. The six-factor AdaBoost model, which included age, IL-6, LDH, SOFA, CO-RADS, and ESR, demonstrated the best performance in predicting mortality risk, and internal validation of the model indicated good predictive performance. Machine learning models can serve as valuable prognostic tools for the early identification of critically ill patients, playing a crucial role in optimizing the allocation of medical resources, guiding patient triage, informing personalized treatment decision-making, and monitoring the progression of COVID-19.\u003c/p\u003e"},{"header":"Declarations","content":"\u003cp\u003e\u003cstrong\u003eEthics approval\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eThe study was approved by the Research Ethics Committee of Zanjan University of Medical Sciences, Zanjan, Iran (IR.ZUMS.REC.1401.204).\u003cstrong\u003e\u0026nbsp;\u003c/strong\u003eThis study was conducted in full accordance with the ethical principles outlined in the Declaration of Helsinki (World Medical Association, 2013)\u003cstrong\u003e.\u0026nbsp;\u003c/strong\u003eInformed consent to participate was obtained from all of the participants\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eClinical trial number\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eNot applicable.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eFunding resources\u0026nbsp;\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eThe authors did not receive any specific grant from funding agencies in the public, commercial, or not-for-profit sectors.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eData availability statement\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eData supporting the findings in this study are immediately available upon reasonable request.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eDeclaration of Competing Interest\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eThe authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eConsent to Publication\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eNot applicable.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eAcknowledgment\u0026nbsp;\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eNot applicable.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eAuthor information\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eNot applicable.\u003c/p\u003e"},{"header":"References","content":"\u003col\u003e\n\u003cli\u003eJiang S, Shi Z, Shu Y, Song J, Gao GF, Tan W, et al. A distinct name is needed for the new coronavirus. Vol. 395, Lancet (London, England). England; 2020. p. 949. \u003c/li\u003e\n\u003cli\u003eOverview WHOR. COVID-19 Epidemiological Update. 2024;(November). \u003c/li\u003e\n\u003cli\u003eContreras S, Priesemann V. Risking further COVID-19 waves despite vaccination. Lancet Infect Dis. 2021 Jun;21(6):745\u0026ndash;6. \u003c/li\u003e\n\u003cli\u003eWang D, Hu B, Hu C, Zhu F, Liu X, Zhang J, et al. Clinical Characteristics of 138 Hospitalized Patients with 2019 Novel Coronavirus-Infected Pneumonia in Wuhan, China. JAMA - J Am Med Assoc. 2020;323(11):1061\u0026ndash;9. \u003c/li\u003e\n\u003cli\u003eGrasselli G, Zangrillo A, Zanella A, Antonelli M, Cabrini L, Castelli A, et al. Baseline Characteristics and Outcomes of 1591 Patients Infected with SARS-CoV-2 Admitted to ICUs of the Lombardy Region, Italy. JAMA - J Am Med Assoc. 2020;323(16):1574\u0026ndash;81. \u003c/li\u003e\n\u003cli\u003eDahai Zhao, Feifei Yao, Lijie Wang, Ling Zheng, Yongjun Gao, Jun Ye, et al. A Comparative Study on the Clinical Features of Coronavirus 2019 (COVID-19) Pneumonia With Other Pneumonias. Clin Infect Dis. 2020;71(15):756\u0026ndash;61. \u003c/li\u003e\n\u003cli\u003eMeyer NJ, Gattinoni L, Calfee CS. Acute respiratory distress syndrome. Lancet (London, England). 2021 Aug;398(10300):622\u0026ndash;37. \u003c/li\u003e\n\u003cli\u003eHsu CY, Lai CC, Yeh YP, Chang-Chuan C, Chen HH. Progression from Pneumonia to ARDS as a Predictor for Fatal COVID-19. J Infect Public Health. 2021 Apr;14(4):504\u0026ndash;7. \u003c/li\u003e\n\u003cli\u003eFeng Z, Yu Q, Yao S, Luo L, Zhou W, Mao X, et al. Early prediction of disease progression in COVID-19 pneumonia patients with chest CT and clinical characteristics. Nat Commun. 2020;11(1):1\u0026ndash;9. \u003c/li\u003e\n\u003cli\u003eHerpe G, Lederlin M, Naudin M, Ohana M, Chaumoitre K, Gregory J, et al. Efficacy of Chest CT for COVID-19 Pneumonia Diagnosis in France. Radiology. 2021 Feb;298(2):E81\u0026ndash;7. \u003c/li\u003e\n\u003cli\u003eShah V, Keniya R, Shridharani A, Punjabi M, Shah J, Mehendale N. Diagnosis of COVID-19 using CT scan images and deep learning techniques. Emerg Radiol. 2021;28(3):497\u0026ndash;505. \u003c/li\u003e\n\u003cli\u003eProkop M, van Everdingen W, van Rees Vellinga T, Quarles van Ufford H, St\u0026ouml;ger L, Beenen L, et al. CO-RADS: A Categorical CT Assessment Scheme for Patients Suspected of Having COVID-19-Definition and Evaluation. Radiology. 2020 Aug;296(2):E97\u0026ndash;104. \u003c/li\u003e\n\u003cli\u003eChen SJ, Chao TF, Chiang MC, Kuo SC, Chen LY, Yin T, et al. Prediction of patient outcome from Acinetobacter baumannii bacteremia with sequential organ failure assessment (SOFA) and acute physiology and chronic health evaluation (APACHE) II scores. Intern Med. 2011;50(8):871\u0026ndash;7. \u003c/li\u003e\n\u003cli\u003eAi Ghozy AR, El Hallage M, Khalid M, Abdalla K. The quick sequential organ failure assessment score for predicting outcome in patients with sepsis and evidence of multiorgan failure at the time of emergency department presentation. Crit Care Shock. 2022;25(3):143\u0026ndash;56. \u003c/li\u003e\n\u003cli\u003ePuhr-Westerheide D, Reich J, Sabel BO, Kunz WG, Fabritius MP, Reidler P, et al. Article sequential organ failure assessment outperforms quantitative chest ct imaging parameters for mortality prediction in covid-19 ards. Diagnostics. 2022;12(1):1\u0026ndash;12. \u003c/li\u003e\n\u003cli\u003eFayed M, Patel N, Angappan S, Nowak K, Vasconcelos F, Penning DH, et al. Sequential Organ Failure Assessment ( SOFA ) Score and Mortality Prediction in Patients With Severe Respiratory Distress Secondary to COVID-. 2022;14(7). \u003c/li\u003e\n\u003cli\u003eLambden S, Laterre PF, Levy MM, Francois B. The SOFA score - Development, utility and challenges of accurate assessment in clinical trials. Crit Care. 2019;23(1):1\u0026ndash;9. \u003c/li\u003e\n\u003cli\u003ePenha D, Pinto EG, Matos F, Hochhegger B, Monaghan C, Taborda-Barata L, et al. CO-RADS: Coronavirus classification review. J Clin Imaging Sci. 2021;11(1):1\u0026ndash;10. \u003c/li\u003e\n\u003cli\u003eEmami A, Javanmardi F, Pirbonyeh N, Akbari A. Prevalence of Underlying Diseases in Hospitalized Patients with COVID-19: a Systematic Review and Meta-Analysis. Arch Acad Emerg Med. 2020;8(1):e35. \u003c/li\u003e\n\u003cli\u003eMehraeen E, Karimi A, Barzegary A, Vahedi F, Afsahi AM, Dadrase O, et al. Since January 2020 Elsevier has created Predictors of mortality in patients with COVID-19\u0026ndash;a systematic review. Eur J Integr Med. 2020;40(January):101226. \u003c/li\u003e\n\u003cli\u003eBanerjee A, Pasea L, Harris S, Gonzalez-Izquierdo A, Torralbo A, Shallcross L, et al. Estimating excess 1-year mortality associated with the COVID-19 pandemic according to underlying conditions and age: a population-based cohort study. Lancet. 2020;395(10238):1715\u0026ndash;25. \u003c/li\u003e\n\u003cli\u003eFigliozzi S, Masci PG, Ahmadi N, Tondi L, Koutli E, Aimo A, et al. Predictors of adverse prognosis in COVID-19: A systematic review and meta-analysis. Eur J Clin Invest. 2020;50(10):1\u0026ndash;15. \u003c/li\u003e\n\u003cli\u003eBonanad C, Garc\u0026iacute;a-blas S. Since January 2020 Elsevier has created a COVID-19 resource centre with free information in English and Mandarin on the novel coronavirus COVID- 19 . The COVID-19 resource centre is hosted on Elsevier Connect , the company \u0026rsquo; s public news and information. Jamda. 2020;(January):915\u0026ndash;8. \u003c/li\u003e\n\u003cli\u003eSimba\u0026ntilde;a-Rivera K, Morocho Jaramillo PR, Velastegui Silva J V., G\u0026oacute;mez-Barreno L, Ventimilla Campoverde AB, Novillo Cevallos JF, et al. High-altitude is associated with better short-term survival in critically ill COVID-19 patients admitted to the ICU. PLoS One. 2022;17(3 March):1\u0026ndash;20. \u003c/li\u003e\n\u003cli\u003eBridgman C, Gerken J, Vincent J, Brooks AE, Zapata I. Revisiting the COVID ‑ 19 fatality rate and altitude association through a comprehensive analysis. Sci Rep [Internet]. 2022;1\u0026ndash;10. Available from: https://doi.org/10.1038/s41598-022-21787-z\u003c/li\u003e\n\u003cli\u003eVahedi A, Tabasi F, Monjazebi F, Hashemian SMR, Tabarsi P, Farzanegan B, et al. Clinical features and outcomes of ICU patients with COVID-19 infection in Tehran, Iran: A single-centered retrospective cohort study. Tanaffos. 2020;19(4):300\u0026ndash;11. \u003c/li\u003e\n\u003cli\u003eBeigmohammadi MT, Amoozadeh L, Rezaei Motlagh F, Rahimi M, Maghsoudloo M, Jafarnejad B, et al. Mortality Predictive Value of APACHE II and SOFA Scores in COVID-19 Patients in the Intensive Care Unit. Can Respir J. 2022;2022. \u003c/li\u003e\n\u003cli\u003eRaith EP, Udy AA, Bailey M, McGloughlin S, MacIsaac C, Bellomo R, et al. Prognostic accuracy of the SOFA score, SIRS criteria, and qSOFA score for in-hospital mortality among adults with suspected infection admitted to the intensive care unit. JAMA - J Am Med Assoc. 2017;317(3):290\u0026ndash;300. \u003c/li\u003e\n\u003cli\u003eEl Aidaoui K, Haoudar A, Khalis M, Kantri A, Ziati J, El Ghanmi A, et al. Predictors of Severity in Covid-19 Patients in Casablanca, Morocco. Cureus. 2020;12(9). \u003c/li\u003e\n\u003cli\u003eLassau N, Ammari S, Chouzenoux E, Gortais H, Herent P, Devilder M, et al. Integrating deep learning CT-scan model, biological and clinical variables to predict severity of COVID-19 patients. Nat Commun. 2021;12(1):1\u0026ndash;11. \u003c/li\u003e\n\u003cli\u003eMoreno R, Rhodes A, Piquilloud L, Hernandez G, Takala J, Gershengorn HB, et al. The Sequential Organ Failure Assessment ( SOFA ) Score : has the time come for an update ? Crit Care. 2023;1\u0026ndash;5. \u003c/li\u003e\n\u003cli\u003eYang Z, Hu Q, Huang F, Xiong S, Sun Y. The prognostic value of the SOFA score in patients with COVID-19. 2021;32(June):1\u0026ndash;9. \u003c/li\u003e\n\u003cli\u003ePawar RD, Shih JA, Balaji L, Grossestreuer A V., Patel P V., Hansen CK, et al. Variation in SOFA (Sequential Organ Failure Assessment) Score Performance in Different Infectious States. J Intensive Care Med. 2021;36(10):1217\u0026ndash;22. \u003c/li\u003e\n\u003cli\u003eWang X, Tang G, Liu Y, Zhang L, Chen B, Han Y, et al. The role of IL-6 in coronavirus, especially in COVID-19. Front Pharmacol. 2022;13(November):1\u0026ndash;12. \u003c/li\u003e\n\u003cli\u003eV\u0026eacute;lez-P\u0026aacute;ez JL, Balde\u0026oacute;n-Rojas L, Ca\u0026ntilde;adas Herrera C, Montalvo MP, Jara FE, Aguayo-Moscoso S, et al. Receiver operating characteristic (ROC) to determine cut-off points of clinical and biomolecular markers to discriminate mortality in severe COVID-19 living at high altitude. BMC Pulm Med [Internet]. 2023;23(1):1\u0026ndash;10. Available from: https://doi.org/10.1186/s12890-023-02691-2\u003c/li\u003e\n\u003cli\u003eLiu X, Wang H, Shi S, Xiao J. Association between IL-6 and severe disease and mortality in COVID-19 disease: a systematic review and meta-analysis. Postgrad Med J. 2021;98(1165):871\u0026ndash;9. \u003c/li\u003e\n\u003cli\u003eHerold T, Jurinovic V, Arnreich C, Lipworth BJ, Hellmuth JC, von Bergwelt-Baildon M, et al. Elevated levels of IL-6 and CRP predict the need for mechanical ventilation in COVID-19. J Allergy Clin Immunol [Internet]. 2020;146(1):128-136.e4. Available from: https://doi.org/10.1016/j.jaci.2020.05.008\u003c/li\u003e\n\u003cli\u003eKomolafe O, Sp P, Br D, Ks G. for the diagnosis of pancreatic necrosis ( Review ). 2017; \u003c/li\u003e\n\u003cli\u003eLi C, Ye J, Chen Q, Hu W, Wang L, Fan Y, et al. Elevated LDH. Definitions. 2020;12(15):15670\u0026ndash;81. \u003c/li\u003e\n\u003cli\u003ePan P, Li Y, Xiao Y, Han B, Su L, Su M, et al. Prognostic assessment of COVID-19 in the intensive care unit by machine learning methods: Model development and validation. J Med Internet Res. 2020;22(11). \u003c/li\u003e\n\u003cli\u003eChen B, Gu HQ, Liu (刘艺) Y, Zhang G, Yang H, Hu H, et al. A model to predict the risk of mortality in severely ill COVID-19 patients. Comput Struct Biotechnol J [Internet]. 2021;19:1694\u0026ndash;700. Available from: https://doi.org/10.1016/j.csbj.2021.03.012\u003c/li\u003e\n\u003c/ol\u003e"}],"fulltextSource":"","fullText":"","funders":[],"hasAdminPriorityOnWorkflow":false,"hasManuscriptDocX":true,"hasOptedInToPreprint":true,"hasPassedJournalQc":"","hasAnyPriority":false,"hideJournal":true,"highlight":"","institution":"","isAcceptedByJournal":false,"isAuthorSuppliedPdf":false,"isDeskRejected":"","isHiddenFromSearch":false,"isInQc":false,"isInWorkflow":false,"isPdf":false,"isPdfUpToDate":true,"isWithdrawnOrRetracted":false,"journal":{"display":true,"email":"[email protected]","identity":"researchsquare","isNatureJournal":false,"hasQc":true,"allowDirectSubmit":true,"externalIdentity":"","sideBox":"","snPcode":"","submissionUrl":"/submission","title":"Research Square","twitterHandle":"researchsquare","acdcEnabled":true,"dfaEnabled":false,"editorialSystem":"","reportingPortfolio":"","inReviewEnabled":false,"inReviewRevisionsEnabled":true},"keywords":"COVID-19, ICU, SOFA score, IL-6, LDH, CO-RADS, mortality prediction models, machine learning","lastPublishedDoi":"10.21203/rs.3.rs-8173098/v1","lastPublishedDoiUrl":"https://doi.org/10.21203/rs.3.rs-8173098/v1","license":{"name":"CC BY 4.0","url":"https://creativecommons.org/licenses/by/4.0/"},"manuscriptAbstract":"\u003ch2\u003eIntroduction:\u003c/h2\u003e\u003cp\u003ePredicting the severity and outcome of COVID-19 is a challenging task. This study investigated the potential of predicting mortality using the SOFA score, CT scan findings based on the CO-RADS system, and biomarkers (including IL-6 and LDH) in intensive care unit (ICU)-admitted patients with COVID-19. Additionally, we developed multivariable models to enhance prognostic accuracy.\u003c/p\u003e\u003ch2\u003eMaterials and methods\u003c/h2\u003e\u003cp\u003eThis retrospective cohort study was conducted on 426 COVID-19 patients admitted to the ICU of a tertiary hospital in Zanjan, Iran, from March to November 2020. The data were collected from patients' medical records. The correlation between variables and mortality was analyzed, and the predictability of mortality was assessed using the receiver operating characteristic (ROC) curve. Cut-off points, sensitivity, and specificity were determined. Conventional logistic regression methods and four machine learning (ML) algorithms were employed to develop mortality prediction models for ICU patients with COVID-19 using Python. The performance of these machine learning models was measured by the area under the receiver operating characteristic curve (AUC). The internal validation of these ML-based models was performed using an integrated 10-fold stratified cross-validation with bootstrap.\u003c/p\u003e\u003ch2\u003eResults\u003c/h2\u003e\u003cp\u003eThe mortality rate was 47.1% (n\u0026thinsp;=\u0026thinsp;200). The mean SOFA (5.23 vs. 3.58, p\u0026thinsp;\u0026lt;\u0026thinsp;0.001) and CO-RADS (5.54 vs. 5.03, p\u0026thinsp;\u0026lt;\u0026thinsp;0.001) scores were significantly higher in the deceased group. Biomarkers were investigated as mortality predictors, and IL-6 (AUC: 0.761, cut-off: \u0026gt;18.95 pg/mL, 65.5% sensitivity, 93.4% specificity) and LDH (AUC: 0.737, cut-off: \u0026gt;437.5 U/L, 63.0% sensitivity, 68.6% specificity) yielded the highest predictability for mortality, followed by SOFA (AUC: 0.701, cut-off: \u0026gt;3, 80.0% sensitivity, 46.5% specificity) and among comorbidities, hypertension and diabetes exhibited significant correlation with mortality risk (p\u0026thinsp;=\u0026thinsp;0.001 and p\u0026thinsp;=\u0026thinsp;0.038, respectively). Using logistic regression methods and four machine learning (ML) algorithms, a six-factor model was developed involving age, IL-6, LDH, SOFA, CO-RADS, and ESR. This model yielded the best AUC in the adaptive boosting (AdaBoost) algorithm (AUC: 0.960, sensitivity: 0.915, specificity: 0.885). According to the internal validation and calibration plots (Fig.\u0026nbsp;3), the closest agreement between predicted and observed probabilities was shown in the categorical boosting (CatBoost) algorithm (Brier score: 0.086, AUC: 0.952).\u003c/p\u003e\u003ch2\u003eConclusions\u003c/h2\u003e\u003cp\u003eIL-6 and LDH showed the highest predictive value for mortality among biomarkers, and the SOFA score showed predictability compared to the CT-based CO-RADS system. The six-factor AdaBoost model showed the best performance in predicting the mortality risk.\u003c/p\u003e","manuscriptTitle":"Machine Learning Prediction Models for COVID-19 ICU Mortality: Model Development and Validation","msid":"","msnumber":"","nonDraftVersions":[{"code":1,"date":"2025-12-03 00:50:06","doi":"10.21203/rs.3.rs-8173098/v1","editorialEvents":[{"type":"communityComments","content":0}],"status":"published","journal":{"display":true,"email":"[email protected]","identity":"researchsquare","isNatureJournal":false,"hasQc":true,"allowDirectSubmit":true,"externalIdentity":"","sideBox":"","snPcode":"","submissionUrl":"/submission","title":"Research Square","twitterHandle":"researchsquare","acdcEnabled":true,"dfaEnabled":false,"editorialSystem":"","reportingPortfolio":"","inReviewEnabled":false,"inReviewRevisionsEnabled":true}}],"origin":"","ownerIdentity":"f31a6c41-d3ea-402b-bf6f-18619de012f0","owner":[],"postedDate":"December 3rd, 2025","published":true,"recentEditorialEvents":[],"rejectedJournal":[],"revision":"","amendment":"","status":"posted","subjectAreas":[],"tags":[],"updatedAt":"2026-01-13T09:55:01+00:00","versionOfRecord":[],"versionCreatedAt":"2025-12-03 00:50:06","video":"","vorDoi":"","vorDoiUrl":"","workflowStages":[]},"version":"v1","identity":"rs-8173098","journalConfig":"researchsquare"},"__N_SSP":true},"page":"/article/[identity]/[[...version]]","query":{"redirect":"/article/rs-8173098","identity":"rs-8173098","version":["v1"]},"buildId":"8U1c8b4HqxoKbykW_rLl7","isFallback":false,"isExperimentalCompile":false,"dynamicIds":[84888],"gssp":true,"scriptLoader":[]}