Skip to main content
NCBI home page
PLOS One logoLink to PLOS One
. 2023 May 19;18(5):e0283469. doi: 10.1371/journal.pone.0283469

Early biochemical analysis of COVID-19 patients helps severity prediction

Andrés Roncancio-Clavijo 1,2,#, Miriam Gorostidi-Aicua 1,#, Ainhoa Alberro 1,2,3, Andrea Iribarren-Lopez 1, Ray Butler 4, Raúl Lopez 4, Jose Antonio Iribarren 5, Diego Clemente 3,4,5,6, Jose María Marimon 7, Javier Basterrechea 8, Bruno Martinez 8, Alvaro Prada 1,2,‡, David Otaegui 1,2,3,‡,*
Editor: Kovy Arteaga-Livias9
PMCID: PMC10198541  PMID: 37205683

Abstract

COVID-19 pandemic has put the protocols and the capacity of our Hospitals to the test. The management of severe patients admitted to the Intensive Care Units has been a challenge for all health systems. To assist in this challenge, various models have been proposed to predict mortality and severity, however, there is no clear consensus for their use. In this work, we took advantage of data obtained from routine blood tests performed on all individuals on the first day of hospitalization. These data has been obtained by standardized cost-effective technique available in all the hospitals. We have analyzed the results of 1082 patients with COVID19 and using artificial intelligence we have generated a predictive model based on data from the first days of admission that predicts the risk of developing severe disease with an AUC = 0.78 and an F1-score = 0.69. Our results show the importance of immature granulocytes and their ratio with Lymphocytes in the disease and present an algorithm based on 5 parameters to identify a severe course. This work highlights the importance of studying routine analytical variables in the early stages of hospital admission and the benefits of applying AI to identify patients who may develop severe disease.

Introduction

COVID-19 is an infectious disease produced by the coronavirus SARS-CoV-2 (Severe Acute Respiratory Syndrome Coronavirus 2). The first outbreak was reported in a seafood market in Wuhan, (Hubei province, China) and the World Health Organization (WHO) declared the situation a public health emergency of international concern on January 30th, 2020 and as a pandemic on March 11th.

COVID-19 disease has a wide clinical spectrum. A relevant proportion of infected individuals is asymptomatic, while others develop from mild symptoms, such as low-grade inflammation, fever and cough, to severe symptoms, that include respiratory failure, metabolic acidosis, coagulopathy, septic shock and multiorgan failure [1]. Although it is a rapidly changing scenario, based on current data, most people infected by the SARS-CoV-2 virus present a mild disease. Besides, depending on the study, different proportions of severe and critical patients, as well as mortality rates have been reported. In the case of Spain, a work that included 4035 hospitalized COVID-19 patients from 127 centres, reported 18.5% intensive care unit (ICU) admission and one-month mortality of 28%, which raised to 42.4% among ICU patients [2]. Prediction of the clinical course of each individual becomes a priority to optimize both the personalized treatment of each patient and the use of the medical resources.

Several works have tried to identify risk and prognostic factors based on demographic, epidemiologic, genetic, and clinical variables [2–6], and several models have been proposed for mortality and severity prediction, but they present diverse outcomes and no consensus has been reached yet on their application [7].

With this in mind, in this work, we took advantage of the routine blood tests performed on all individuals on the first day of hospitalization and analyzed the results of 1,082 COVID-19 patients generating a predictive model, based on these first-day data, that allows us to efficiently predict the course of the disease. Our results also highlight the importance of the immature granulocytes and their ratio with lymphocytes as parameters to be studied.

Methods

This study has been conducted at the Donostia University Hospital, which covers the region of Gipuzkoa (Basque Country, Spain), with a population of 720,000 inhabitants. Data of the first blood analysis after the COVID-19 diagnosis have been retrieved from the clinical records. The inclusion criteria were having a SARS-CoV-2 positive PCR between February and April 2020 and being admitted to the hospital. Demographic data and variables are presented in Tables 1 and 2. The data has been obtained fully anonymized. The data has been obtained anonymously and the follow-up has been done until 30 days after the discharge of the hospital. This retrospective chart review study involving human participants was in accordance with the ethical standards of the institutional and national research committee and with the 1964 Helsinki Declaration and its later amendments or comparable ethical standards. The Human Investigation Committee (IRB) of Hospital Universitario Donostia (PI2020076) approved this study.

Table 1. Demographic and clinical data of the study cohort.

Percentages for the first column were calculated for the total number of patients (1082). The percentages shown for Females and Males were calculated with respect to the total number of patients in each of the conditions, and the p-value was calculated for these comparisons. IQR = Interquantile range. ICU = Intensive care Unit.

  All Female Male p-value
Samples (%) 1082 545 (50.4%) 537 (49.6%) 0.81
Mean age (min-max) 66.5 (10–99) 66.5 (10–99) 66.5 (17–95) --
Severe cases (%) 449 (41.5%) 184 (41%) 265 (59%) < 0.001 ***
ICU admission (%) 60 (5.6%) 16 (26.7%) 44 (73.3%) < 0.001 ***
Deceased (%) 140 (12.9%) 54 (38.6%) 86 (61.4%) 0.004 **
Open in a new tab

Table 2. The normal range of the study variables and the data from COVID-19 patients were compared.

Fold change was calculated when the median value of patients was above the maximum or below the minimum value of the normal range. IQR = Interquantile range.

  Normal Range COVID-19 (min-max) Fold Change
Albumin (g/dl) 3.7–5.1 3.9 (2.07–4.94) -
Creatine (mg/dl) 0.4–1 1.02 (0.27–13.5) 1.02
Urea (mg/dl) 10–65 43.97 (8–201) -
Triglycerides (mg/dl) 35–135 129.04 (34–1265) -
Creatine kinase (CK) (U/l) 0–188 150.92 (20–2662) -
ALT/SGPT (%) 0–33 25.79 (2–395.5) -
AST/SGOT (U/l) 0–31 30.33 (7–503) -
LDH (U/l) 60–190 277 (110–2290) 1.46
Troponin T (TnT) (ng/l) 0–14 20.32 (2–393) 1.45
C-reactive protein (mg/l) 0–5 50.16 (0–410) 10.03
IL-6 (pg/ml) 0–7 25.64 (1.72–166) 3.66
Prothrombin Index (%) 70–140 93.58 (4–137) -
Fibrinogen (mg/dl) 200–400 530.15 (266–780) 1.33
D-dimer (ng/ml) 0–500 1351.25 (180–21000) 2.7
Ferritin (ng/ml) 15–150 426.71 (6.7–4557.3) 2.84
Procalcitonin (ng/ml) 0–0.5 0.36 (0.01–18.51) -
Lactate (mmol/l) 0.3–2 1.37 (0.5–6) -
Platelets (103/μl) 140–400 215.69 (12–729) -
Leukocytes (103/μl) 3.8–10 7.07 (0.03–77.44) -
Neutrophils (103/μl) 1.6–7.5 4.5 (0–18.81) -
Neutrophils (%) 40–75 63.16 (0–95.6) -
Eosinophils (103/μl) 0–0.6 0.12 (0–1.85) -
Eosinophils (%) 0.5–7 1.75 (0–16.9) -
Basophils (103/μl) 0–0.2 0.03 (0–0.16) -
Basophils (%) 0–1.5 0.45 (0–2.1) -
Immature granulocytes (103/μl) 0–0.03 0.04 (0–0.97) 1.35
Immature granulocytes (%) 0–0.5 0.55 (0–23.2) 1.11
Lymphocytes (103/μl) 0.9–3.5 1.82 (0.01–65.51) -
Lymphocytes (%) 19–48 25.33 (0.9–88.9) -
Monocytes (103/μl) 0.2–0.9 0.58 (0.02–7.36) -
Monocytes (%) 3.05–12 8.96 (0.5–66.7) -
Partial pressure of oxygen (mmHg) 83–108 73.18 (14–271) -1.13
Oxygen saturation (%) 95–99 92.94 (21.2–100.8) -1.02
Open in a new tab

The severity categorization was done a posteriori according to [8]. “Severe cases” were defined as the ones meeting any of the following criteria: (1) Respiratory distress (≧30 breaths/min); (2) Oxygen saturation≤93% at rest; (3) Arterial partial pressure of oxygen (PaO2)/fraction of inspired oxygen (FiO2)≦300mmHg. Those patients that did not meet any of the conditions were considered “Non-severe cases”.

Continuous variables have been described by the median, minimum and maximum and interquartile range while categorical variables have been described by the number of cases and percentages. For the continuous variables, the comparison of the means has been done using T-test, when the variables were normally distributed, and Mann-Whitney-Wilcoxon (2 categories) / Kruskal-Wallis (>2 categories) tests for the non-normally distributed. For the categorical variables, Chi-squared and Cramér’s V index test were applied. The fold change (FC) was calculated for the variables that had a mean value out of the normal range. For increased variables, the mean value was divided by the maximum value of the normal range, and in those below the normal range, by dividing the minimum value of the normal range by the mean value. FC values >|2| and a p-value lower than 0.05 were considered relevant. All statistical analyses have been done using R (R.3.6.3) [9]. In the charts, p-values has been explained as follows: * → p<0.05, ** → p<0.01 and *** →p<0.001.

We used AutoDiscovery (Butler Scientifics) for a stratified feature selection and to generate our final classification model. Briefly, a data set of 42 features with the data of 1,082 patients was constructed and 37 of these variables were chosen for the analysis [33 variables identified plus age group, sex, severity and the Lymphocyte to Immature Granulocyte ratio (L/IG)]. This L/IG ratio has been included based in clinical observation and in the analysis of our data. The next step was an automated data exploration, where the association and correlation between the variables were analyzed (AutoDiscovery Strength Score) to avoid model overfitting and to identify the patient’s features that were more associated with the predictive goal. In this specific case, the association of the variables with the severity variable was evaluated and one of the correlated pair of variables was removed, if any. To do that, AutoDiscovery selected the proper numerical method based on the data type and distribution of the variables assessed. Numerical tests applied in each case are Spearman’s Rank Correlation, Variance Analysis (ANOVA one-way, U Mann-Whitney, and Kruskal-Wallis; normality was tested in these cases with D’Agostino/Pearson methods), and Cramer’s V Contingency Index.

This massive data exploration process was carried out systematically for each patient group (stratum), considering only those correlations evaluated in a data subset with n≥5, two-tailed p < 0.05 and Spearman’s rank correlation coefficient > 0.8 to be relevant.

Given the nature of this multiple testing method, a high significance threshold was calculated based on Benjamini-Hochberg method (False-Discovery Rate) to classify the rest associations.

The Feature Selection process, where, based on Recursive Feature Elimination (RFE) and the AutoDiscovery Strength Score, age decade and the Lymphocytes to Immature Granulocyte ratio were added as variables for the model.

The model was trained and evaluated using three models: Top-5, Top-10 and Top-18 (using the 5,10 and 18 most significant variables), plus Age group and L/IG). The training and evaluation processes included the utilization of multi-class classifiers, such as Decision Trees, Logistic Regression, Naïve Bayes, K-Neighbors, SVM and Random Forest. The Model Training step is described in Fig 1, using the 60% of the samples for training, a 20% for validating and the final 20% for the testing, where the most optimized accuracy is obtained. Finally, a Cross-Validation method (K-fold, K = 10) is applied to decrease overfitting impact. Sensitivity, Specificity and accuracy has been calculated trough SciPy [10]. F1-score has been calculated from the Precision and the recall.

Fig 1. Schematic representation of the classifier analysis.

Fig 1

Open in a new tab

The results of all these steps are the Confusion Matrix, the accuracy and the ROC curve of each of the three Top models (S1 File).

Results

The sex, age and clinical data of the 1082 participants are presented in Table 1. The incidence of severe cases, ICU admission and deceased patients was significantly higher among male individuals. A higher percentage of severe cases was reported with increasing age, while it is reduced in nonagenarians when compared to octogenarians (p-value = 0.04) (Fig 2).

Fig 2. Age distribution of COVID-19 severity.

Fig 2

Open in a new tab

For each of the age ranges the percentage of Non-severe and Severe cases are shown. Age is presented in years.

The median value of most of the variables was in the normal range, only 4 of them have a fold change > |2|. Patients have more D-dimer, Ferritin, IL-6, and remarkably, 10.03 times more C-reactive proteins (Table 2).

To determine if any of these variables could be good predictors of disease severity, patients were grouped based on their severity outcome. Notably, as shown in Table 3, we found that 32 out of the 33 variables tested were significantly different between severe and non-severe cases. Eleven variables show lower values in severe group, 21 were increased. There was a strong statistical significance for most of the variables, with 26 variables with a p-value < 0.001. We also calculated the L/IG for all the patients. The obtained results demonstrate that there is a strongly significant reduction of the L/IG in the first blood analysis of patients that then develop a severe disease.

Table 3. Comparison of the study variables between the COVID-19 patients that have a “Non-severe course” and the ones that develop a “Severe disease”.

32 of the 33 evaluated variables are significantly different. For an easier visual interpretation of the results, analytes with lower values in severe patients are shown in white and the analytes with higher values in gray. The “Sample n” column indicates the number of data available for each analyte. IQR = Interquantile range.

  Normal Range COVID-19 (min-max) Non-Severe (min-max) Severe (min-max) p-value Sample n
Albumin (g/dl) 3.7–5.1 3.9 (2.07–4.94) 4 (2.07–4.94) 3.76 (2.15–4.82) < 0.001 *** 309
Creatine (mg/dl) 0.4–1 1.02 (0.27–13.5) 0.92 (0.27–13.5) 1.16 (0.29–10.44) < 0.001 *** 869
Urea (mg/dl) 10–65 43.97 (8–201) 38.49 (8–193.2) 51.21 (11–201) < 0.001 *** 685
Triglycerides (mg/dl) 35–135 129.04 (34–1265) 118.18 (34–492) 156.74 (54–1265) < 0.001 *** 213
Creatine kinase (CK) (U/l) 0–188 150.92 (20–2662) 137.78 (25–2662) 168.97 (20–1322) 0.03 * 216
ALT/SGPT (%) 0–33 25.79 (2–395.5) 24.9 (5–395.5) 27.32 (2–228.5) 0.01 ** 691
AST/SGOT (U/l) 0–31 30.33 (7–503) 29.79 (7–503) 31.26 (9–115) < 0.001 *** 462
LDH (U/l) 60–190 277 (110–2290) 245.48 (117–727) 319.81 (110–2290) < 0.001 *** 382
Troponin T (TnT) (ng/l) 0–14 20.32 (2–393) 11.55 (2–162) 32.79 (2–393) < 0.001 *** 213
C-reactive protein (mg/l) 0–5 50.16 (0–410) 31.89 (0–206.1) 74.26 (0.2–410) < 0.001 *** 603
IL-6 (pg/ml) 0–7 25.64 (1.72–166) 7.17 (1.72–25.6) 99.5 (33–166) 0.04 * 10
Prothrombin Index (%) 70–140 93.58 (4–137) 96.45 (6–136) 89.72 (4–137) 0.02 * 539
Fibrinogen (mg/dl) 200–400 530.15 (266–780) 473.67 (266–780) 600.75 (426–765) 0.01 ** 27
D-dimer (ng/ml) 0–500 1351.25 (180–21000) 861.93 (180–12650) 2010.07 (270–21000) < 0.001 *** 359
Ferritin (ng/ml) 15–150 426.71 (6.7–4557.3) 317.18 (6.7–2100) 629.86 (20.4–4557.3) < 0.001 *** 354
Procalcitonin (ng/ml) 0–0.5 0.36 (0.01–18.51) 0.17 (0.01–16.8) 0.59 (0.02–18.51) < 0.001 *** 349
Lactate (mmol/l) 0.3–2 1.37 (0.5–6) 1.19 (0.5–3.9) 1.5 (0.6–6) < 0.001 *** 268
Platelets (103/μl) 140–400 215.69 (12–729) 221.79 (44–729) 206.29 (12–560) < 0.001 *** 874
Leukocytes (103/μl) 3.8–10 7.07 (0.03–77.44) 6.56 (1.41–54.32) 7.86 (0.03–77.44) < 0.001 *** 875
Neutrophils (103/μl) 1.6–7.5 4.5 (0–18.81) 4.01 (0.61–14.21) 5.25 (0–18.81) < 0.001 *** 875
Neutrophils (%) 40–75 63.16 (0–95.6) 60.28 (7.3–90.5) 67.59 (0–95.6) < 0.001 *** 875
Eosinophils (103/μl) 0–0.6 0.12 (0–1.85) 0.13 (0–1.85) 0.1 (0–1.36) < 0.001 *** 875
Eosinophils (%) 0.5–7 1.75 (0–16.9) 2.01 (0–16.9) 1.35 (0–14.6) < 0.001 *** 875
Basophils (103/μl) 0–0.2 0.03 (0–0.16) 0.03 (0–0.13) 0.03 (0–0.16) 0.02 * 875
Basophils (%) 0–1.5 0.45 (0–2.1) 0.48 (0–2.1) 0.4 (0–1.8) < 0.001 *** 875
Immature granulocytes (103/μl) 0–0.03 0.04 (0–0.97) 0.03 (0–0.29) 0.06 (0–0.97) < 0.001 *** 875
Immature granulocytes (%) 0–0.5 0.55 (0–23.2) 0.42 (0–8.5) 0.75 (0–23.2) < 0.001 *** 875
Lymphocytes (103/μl) 0.9–3.5 1.82 (0.01–65.51) 1.79 (0.26–48.29) 1.86 (0.01–65.51) < 0.001 *** 875
Lymphocytes (%) 19–48 25.33 (0.9–88.9) 27.77 (3.5–88.9) 21.58 (0.9–86) < 0.001 *** 875
Monocytes (103/μl) 0.2–0.9 0.58 (0.02–7.36) 0.57 (0.12–1.9) 0.6 (0.02–7.36) - 875
Monocytes (%) 3.05–12 8.96 (0.5–66.7) 9.18 (1.7–38.6) 8.63 (0.5–66.7) < 0.001 *** 875
Partial pressure of oxygen (mmHg) 83–108 73.18 (14–271) 81.73 (27–154) 68.53 (14–271) < 0.001 *** 321
Oxygen saturation (%) 95–99 92.94 (21.2–100.8) 96.66 (46.5–99.9) 90.92 (21.2–100.8) < 0.001 *** 318
Lymphocyte to Immature Granulocyte ratio   84.97 (0.75–889) 101.26 (2.75–889) 59.94 (0.75–430) < 0.001 *** 875
Open in a new tab

Since almost all variables appeared significantly different between severity groups, even if the mean value of most of them was in the normal range, it was decided to create a classification model to find the most relevant variables that would be able to predict the severity outcome of the patients at an early stage of the disease.

AutoDiscovery software was used for the creation of a predictive model. In the Automated Data Exploration step an association of almost all variables with severity variable was found. The ranking of the variables obtained in the Feature Selection step is shown in Fig 3, where the specific variables of each model are described.

Fig 3. List of the variables of the three classifications models.

Fig 3

Open in a new tab

Among all the multi-class classifiers, Random Forest showed the best performance. After the Cross-Validation process and the application of RFE, the results have shown an accuracy of around 60% for the three models, being the Top-5 the best model, with an accuracy of 67.33%. The Top-5 model also achieved the best confusion matrix, the best F1-score (0.68 for severe and 0.70 for Non-severe) and ROC curve (Fig 4).

Fig 4. Classification models for the prediction of disease severity.

Fig 4

Open in a new tab

The confusion matrix and ROC curves for the (A) Top-5, (B) Top-10 and (C) Top-18 models are shown.

Discussion

Our study showed higher severity, ICU admission and death rates among male patients (Table 1), in accordance with previously reports [11]. The age distribution of severe cases followed a very similar pattern to that reported in the Italian population, with an increasing severity with age and a maximum peak of severe cases in octogenarians, followed by a decrease of this percentage in nonagenarians. Besides, the ICU occupancy (critical patients) was also similar, resulting in a 5.5% in our analysis and a 5.3% in the Italian study. The mortality rate was also analogous to previous reports, which was 12.94% in the present study, close to the 11% estimated for the Spanish population in the literature [2].

From the studied variables, only the absolute number of monocytes did not show significant differences between groups (Table 3). The analytical alterations observed in the biochemistry of COVID-19 patients are the reflection of the degree and the severity of multi-organ involvement [12], similar to what is observed in septic conditions of bacterial origin. These conditions commonly appear due to endothelial dysfunction and abnormal activation of the coagulation cascade, whose origins and mechanism are not completely understood yet [13].

Our results highlight the importance of use AI algorithms that offer the opportunity to identify new association that could be used as good predictors even when they are within normal ranges.

As reported in COVID-19, the cytokine storm is largely responsible for disease severity and it can have lethal outcomes [14]. Accordingly to previous reports, in our cohort, the percentage of lymphocytes showed statistically significant lower levels in severe compared to non-severe patients. Other parameters such as platelets, leukocytes, neutrophils, eosinophils, basophils and the percentage of monocytes showed statistically significant differences between severe and non-severe patients. These results are in accordance with what is reported in the literature [15]. Machine learning approaches have proposed as important variables in COVID19 severity prediction hyperferritinaemia, hypocalcaemia, hypoxia, hypoxemia and high levels of lactate dehydrogenase among others [16] and the use of metabolites as more powerful biochemical variables [17]

A parameter that was initially little explored in COVID-19 disease and that has shown to have prognostic implications is the number of immature granulocytes (IG) [18, 19], suggesting that the dysregulation of the myeloid compartment plays a role in the physiopathology of the COVID-19 disease. IG is the variable that show the greatest discriminatory weight, between those chosen by the AI algorithm. The percentage of immature granulocytes give us an approach to the presence of a higher production of myeloid-derived immature cells with suppression capacity (MDSCs). Interestingly, augmented MDSCs have been found in COVID-19 patients with a severe disease course [20–22], and it has been proposed that MDSCs expand their numbers under inflammatory stimuli [23] and that they can explain, at least in part, processes such as L-arginine mediated lymphopenia [24].

Therefore, the observed inverse relation between lymphocytes and immature myeloid cells, the positive relation between absolute numbers of IG and IL-6, and the physiopathological basis that explains, in part, the lymphopenia, led us to calculate the Lymphocyte to Immature Granulocyte ratio. Importantly, we found that this indirect measure of systemic inflammation was significantly decreased in patients that developed a severe COVID-19 disease when compared to the ones that only suffered a non-severe disease (Table 3). In line with our results, other authors found the immature neutrophil-to-VD2 (or CD8) T-cell ratio as an early marker for severe COVID-19 disease, predicting pneumonia and hypoxia onset. Moreover, it has been also reported, among other variables, a differential immature granulocyte to lymphocyte ratio between critical and non-critical COVID-19 patients, reinforcing the relevance of this measure [25].

The model with just 5 variables shows an AUC = 0.78 and a F1-score = 0.69. that is near to some of the previously presented AUC to predict severity [16, 17] showing promising utility as a clinical tool for predicting disease severity.

In conclusion, the present work highlights the importance of the study of analytic variables in the early phases of hospitalization by an easy and standard cost-effective technique present in all the hospitals. Our results also underline the benefits of applying AI methods to identify the best variables, even although they present values in normal ranges, to identify the best classifier. The prompt identification of the risk group is essential to thoroughly follow them and implement the necessary analytical and clinical interventions.

Supporting information

S1 File. Supporting information, including one figure and raw data.

(ZIP)

Click here for additional data file. (180KB, zip)

Data Availability

All relevant data are within the paper and its Supporting Information files.

Funding Statement

This Project has been funded by Instituto de Salud Carlos III (COV20/00314) that include part of FEDER founds. AA is funded by Basque government through a post-doc Grant. The funders had no role in study design, data collection and analysis, decision to publish or preparation of the manuscript."

References

  • 1.Cascella M, Rajnik M, Cuomo A, Dulebohn SC, Di Napoli R. Features, Evaluation, and Treatment of Coronavirus. 2020. [PubMed] [Google Scholar]
  • 2.Berenguer J, Ryan P, Rodríguez-Baño J, Jarrín I, Carratalà J, Pachón J, et al. Characteristics and predictors of death among 4035 consecutively hospitalized patients with COVID-19 in Spain. Clin Microbiol Infect 2020;26:1525–36. doi: 10.1016/j.cmi.2020.07.024 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 3.Zhou F, Yu T, Du R, Fan G, Liu Y, Liu Z, et al. Clinical course and risk factors for mortality of adult inpatients with COVID-19 in Wuhan, China: a retrospective cohort study. Lancet 2020;395:1054–62. doi: 10.1016/S0140-6736(20)30566-3 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 4.Bonetti G, Manelli F, Patroni A, Bettinardi A, Borrelli G, Fiordalisi G, et al. Laboratory predictors of death from coronavirus disease 2019 (COVID-19) in the area of Valcamonica, Italy. Clin Chem Lab Med 2020;58:1100–5. doi: 10.1515/cclm-2020-0459 [DOI] [PubMed] [Google Scholar]
  • 5.Van Dam PMEL, Zelis N, Stassen P, Van Twist DJL, De Leeuw PW, Van Kuijk S, et al. Validating the RISE up score for predicting prognosis in patients with COVID-19 in the emergency department: A retrospective study. BMJ Open 2021;11:45141. 10.1136/bmjopen-2020-045141. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 6.Gupta RK, Harrison EM, Ho A, Docherty AB, Knight SR, van Smeden M, et al. Development and validation of the ISARIC 4C Deterioration model for adults hospitalised with COVID-19: a prospective cohort study. Lancet Respir Med 2021;9:349–59. doi: 10.1016/S2213-2600(20)30559-2 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 7.Wynants L, Van Calster B, Collins GS, Riley RD, Heinze G, Schuit E, et al. Prediction models for diagnosis and prognosis of covid-19: Systematic review and critical appraisal. BMJ 2020;369:26. doi: 10.1136/bmj.m1328 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 8.On NHC& SA of TCM. Diagnosis and Treatment Protocol for Novel Coronavirus Pneumoniae. (trial version 7). 2020. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 9.Team RC. R: A language and environment for statistical computingitle 2017. [Google Scholar]
  • 10.Virtanen P, Gommers R, Oliphant TE, Haberland M, Reddy T, Cournapeau D, et al. SciPy 1.0: fundamental algorithms for scientific computing in Python. Nat Methods 2020;17:261–72. doi: 10.1038/s41592-019-0686-2 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 11.Karagiannidis C, Mostert C, Hentschker C, Voshaar T, Malzahn J, Schillinger G, et al. Case characteristics, resource use, and outcomes of 10 021 patients with COVID-19 admitted to 920 German hospitals: an observational study. Lancet Respir Med 2020;8:853–62. 10.1016/S2213-2600(20)30316-7. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 12.Zaim S, Chong JH, Sankaranarayanan V, Harky A. COVID-19 and Multiorgan Response. Curr Probl Cardiol 2020;45. doi: 10.1016/j.cpcardiol.2020.100618 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 13.Lazzaroni MG, Piantoni S, Masneri S, Garrafa E, Martini G, Tincani A, et al. Coagulation dysfunction in COVID-19: The interplay between inflammation, viral infection and the coagulation system. Blood Rev 2021;46. doi: 10.1016/j.blre.2020.100745 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 14.Tang Y, Liu J, Zhang D, Xu Z, Ji J, Wen C. Cytokine Storm in COVID-19: The Current Evidence and Treatment Strategies. Front Immunol 2020;11:1708. doi: 10.3389/fimmu.2020.01708 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 15.Zhu Z, Cai T, Fan L, Lou K, Hua X, Huang Z, et al. Clinical value of immune-inflammatory parameters to assess the severity of coronavirus disease 2019. Int J Infect Dis 2020;95:332–9. doi: 10.1016/j.ijid.2020.04.041 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 16.Cobre A de F, Stremel DP, Noleto GR, Fachi MM, Surek M, Wiens A, et al. Diagnosis and prediction of COVID-19 severity: can biochemical tests and machine learning be used as prognostic indicators? Comput Biol Med 2021;134. doi: 10.1016/j.compbiomed.2021.104531 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 17.de Fátima Cobre A, Surek M, Stremel DP, Fachi MM, Lobo Borba HH, Tonin FS, et al. Diagnosis and prognosis of COVID-19 employing analysis of patients’ plasma and serum via LC-MS and machine learning. Comput Biol Med 2022;146. doi: 10.1016/j.compbiomed.2022.105659 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 18.Mare TA, Treacher DF, Shankar-Hari M, Beale R, Lewis SM, Chambers DJ, et al. The diagnostic and prognostic significance of monitoring blood levels of immature neutrophils in patients with systemic inflammation. Crit Care 2015;19. 10.1186/s13054-015-0778-z. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 19.Schulte-Schrepping J, Reusch N, Paclik D, Baßler K, Schlickeiser S, Zhang B, et al. Severe COVID-19 Is Marked by a Dysregulated Myeloid Cell Compartment. Cell 2020;182:1419–1440.e23. doi: 10.1016/j.cell.2020.08.001 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 20.Schulte-Schrepping J, Reusch N, Paclik D, Baßler K, Schlickeiser S, Zhang B, et al. Suppressive myeloid cells are a hallmark of severe COVID-19. MedRxiv 2020;19:2020.06.03. 10.1007/BF03180576 . [DOI] [Google Scholar]
  • 21.Carissimo G, Xu W, Kwok I, Abdad MY, Chan YH, Fong SW, et al. Whole blood immunophenotyping uncovers immature neutrophil-to-VD2 T-cell ratio as an early marker for severe COVID-19. Nat Commun 2020;11:1–12. 10.1038/s41467-020-19080-6. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 22.Zini G, Bellesi S, Ramundo F, d’Onofrio G. Morphological anomalies of circulating blood cells in COVID ‐19. Am J Hematol 2020;95:870–2. doi: 10.1002/ajh.25824 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 23.Calahorra L, Camacho-Toledano C, Serrano-Regal MP, Ortega MC, Clemente D. Regulatory Cells in Multiple Sclerosis: From Blood to Brain. Biomedicines 2022;10. 10.3390/BIOMEDICINES10020335. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 24.Bronte V, Zanovello P. Regulation of immune responses by L-arginine metabolism. Nat Rev Immunol 2005;5:641–54. doi: 10.1038/nri1668 [DOI] [PubMed] [Google Scholar]
  • 25.Linssen J, Ermens A, Berrevoets M, Seghezzi M, Previtali G, Brugge S van der S van der, et al. A novel haemocytometric covid-19 prognostic score developed and validated in an observational multicentre european hospital-based study. Elife 2020;9:1–37. 10.7554/eLife.63195. [DOI] [PMC free article] [PubMed] [Google Scholar]

Decision Letter 0

Kovy Arteaga-Livias

17 Aug 2022

PONE-D-22-16300Early biochemical analysis of COVID-19 patients helps severity predictionPLOS ONE

Dear Dr. Otaegui,

Thank you for submitting your manuscript to PLOS ONE. After careful consideration, we feel that it has merit but does not fully meet PLOS ONE’s publication criteria as it currently stands. Therefore, we invite you to submit a revised version of the manuscript that addresses the points raised during the review process. We have seen citation suggestions in the reviews. Remind you that this is not mandatory and that the inclusion or not of citations will have no influence on the decision made about your manuscript.

Please submit your revised manuscript by Oct 01 2022 11:59PM. If you will need more time than this to complete your revisions, please reply to this message or contact the journal office at plosone@plos.org. When you're ready to submit your revision, log on to https://www.editorialmanager.com/pone/ and select the 'Submissions Needing Revision' folder to locate your manuscript file.

Please include the following items when submitting your revised manuscript:

  • A rebuttal letter that responds to each point raised by the academic editor and reviewer(s). You should upload this letter as a separate file labeled 'Response to Reviewers'.

  • A marked-up copy of your manuscript that highlights changes made to the original version. You should upload this as a separate file labeled 'Revised Manuscript with Track Changes'.

  • An unmarked version of your revised paper without tracked changes. You should upload this as a separate file labeled 'Manuscript'.

If you would like to make changes to your financial disclosure, please include your updated statement in your cover letter. Guidelines for resubmitting your figure files are available below the reviewer comments at the end of this letter.

If applicable, we recommend that you deposit your laboratory protocols in protocols.io to enhance the reproducibility of your results. Protocols.io assigns your protocol its own identifier (DOI) so that it can be cited independently in the future. For instructions see: https://journals.plos.org/plosone/s/submission-guidelines#loc-laboratory-protocols. Additionally, PLOS ONE offers an option for publishing peer-reviewed Lab Protocol articles, which describe protocols hosted on protocols.io. Read more information on sharing protocols at https://plos.org/protocols?utm_medium=editorial-email&utm_source=authorletters&utm_campaign=protocols.

We look forward to receiving your revised manuscript.

Kind regards,

Kovy Arteaga-Livias

Academic Editor

PLOS ONE

Journal Requirements:

When submitting your revision, we need you to address these additional requirements.

1. Please ensure that your manuscript meets PLOS ONE's style requirements, including those for file naming. The PLOS ONE style templates can be found at

https://journals.plos.org/plosone/s/file?id=wjVg/PLOSOne_formatting_sample_main_body.pdf and

https://journals.plos.org/plosone/s/file?id=ba62/PLOSOne_formatting_sample_title_authors_affiliations.pdf

2. Thank you for stating the following financial disclosure:

"The funders had no role in study design, data collection and analysis, decision to publish, or preparation of the manuscript"

At this time, please address the following queries:

a)        Please clarify the sources of funding (financial or material support) for your study. List the grants or organizations that supported your study, including funding received from your institution.

b)        State what role the funders took in the study. If the funders had no role in your study, please state: “The funders had no role in study design, data collection and analysis, decision to publish, or preparation of the manuscript.”

c)        If any authors received a salary from any of your funders, please state which authors and which funders.

d)        If you did not receive any funding for this study, please state: “The authors received no specific funding for this work.”

Please include your amended statements within your cover letter; we will change the online submission form on your behalf.

3. We note that you have indicated that data from this study are available upon request. PLOS only allows data to be available upon request if there are legal or ethical restrictions on sharing data publicly. For more information on unacceptable data access restrictions, please see http://journals.plos.org/plosone/s/data-availability#loc-unacceptable-data-access-restrictions.

In your revised cover letter, please address the following prompts:

a) If there are ethical or legal restrictions on sharing a de-identified data set, please explain them in detail (e.g., data contain potentially sensitive information, data are owned by a third-party organization, etc.) and who has imposed them (e.g., an ethics committee). Please also provide contact information for a data access committee, ethics committee, or other institutional body to which data requests may be sent.

b) If there are no restrictions, please upload the minimal anonymized data set necessary to replicate your study findings as either Supporting Information files or to a stable, public repository and provide us with the relevant URLs, DOIs, or accession numbers. For a list of acceptable repositories, please see http://journals.plos.org/plosone/s/data-availability#loc-recommended-repositories.

We will update your Data Availability statement on your behalf to reflect the information you provide.

4. We note that you have stated that you will provide repository information for your data at acceptance. Should your manuscript be accepted for publication, we will hold it until you provide the relevant accession numbers or DOIs necessary to access your data. If you wish to make changes to your Data Availability statement, please describe these changes in your cover letter and we will update your Data Availability statement to reflect the information you provide.

5. Please ensure that you include a title page within your main document. You should list all authors and all affiliations as per our author instructions and clearly indicate the corresponding author.

6. Your ethics statement should only appear in the Methods section of your manuscript. If your ethics statement is written in any section besides the Methods, please delete it from any other section.

[Note: HTML markup is below. Please do not edit.]

Reviewers' comments:

Reviewer's Responses to Questions

Comments to the Author

1. Is the manuscript technically sound, and do the data support the conclusions?

The manuscript must describe a technically sound piece of scientific research with data that supports the conclusions. Experiments must have been conducted rigorously, with appropriate controls, replication, and sample sizes. The conclusions must be drawn appropriately based on the data presented.

Reviewer #1: Yes

Reviewer #2: Partly

**********

2. Has the statistical analysis been performed appropriately and rigorously?

Reviewer #1: Yes

Reviewer #2: No

**********

3. Have the authors made all data underlying the findings in their manuscript fully available?

The PLOS Data policy requires authors to make all data underlying the findings described in their manuscript fully available without restriction, with rare exception (please refer to the Data Availability Statement in the manuscript PDF file). The data should be provided as part of the manuscript or its supporting information, or deposited to a public repository. For example, in addition to summary statistics, the data points behind means, medians and variance measures should be available. If there are restrictions on publicly sharing data—e.g. participant privacy or use of data from a third party—those must be specified.

Reviewer #1: Yes

Reviewer #2: Yes

**********

4. Is the manuscript presented in an intelligible fashion and written in standard English?

PLOS ONE does not copyedit accepted manuscripts, so the language in submitted articles must be clear, correct, and unambiguous. Any typographical or grammatical errors should be corrected at revision, so please note any specific errors here.

Reviewer #1: Yes

Reviewer #2: No

**********

5. Review Comments to the Author

Please use the space provided to explain your answers to the questions above. You may also include additional comments for the author, including concerns about dual publication, research ethics, or publication ethics. (Please upload your review as an attachment if it exceeds 20,000 characters)

Reviewer #1: Overall, the study was very well conducted. I start by congratulating the authors. Below are my comments and I hope they are useful to improve your study.

1. Line 63, “Continuous variables have been described by the mean, minimum and maximum data”. Clinical data have almost always been shown not to have a normal distribution, my suggestion is that these results be presented in the form of median and interquartile range.

2. Line 73, “Statistical significance was defined as follows: p<0.001***”. Why are the levels of significance different? Was there any sampling procedure that was done to get to this point?

3.On the lines 99-100, "The Model Training step is 100 described in Figure 1". I saw to figure 1, it's actually just a graph of the age distribution as a function of severity. I understood nothing. Are you sure it's figure 1?

4.Lines 100-101, “Finally, a Cross-Validation 102 method is applied to decrease overfitting impact”. What type of cross-validation was used? Venetian blind?, leave-one-out?

5.I encourage the authors to insert the formulas for calculating the metrics for evaluating the performance of the models: sensitivity, specificity, accuracy(material and methods). Still on the evaluation of the performance of the models, I ask that the authors discuss the following scientific articles: PMID: 35751188, PMID: 34091385.

6.Still on the evaluation of the performance of the models, I ask that authors present a table of confusion matrix with: true positives, true negatives, false positives, false negatives, sensitivity, specificity and accuracy. This must be shown for all classifiers used. For more understanding, see the PMID paper table: 35751188.

7.What types of preprocessing are used for machine learning data? I encourage authors to clarify, in the material and methods section and in the Results section.

Reviewer #2: The article needs revision and re-writing with good language and grammar. 

1. What is the novelty of this research?

2.What did the authors find new ?

3. The data is old and dates back to the year 2020, and much research has been published on this topic.

4. Why did the authors not focus on albumin levels, despite the fact that albumin was found to decrease in much of the research

5.Why did the authors not focus on LDH levels, despite the fact that albumin was found to increase in much of the research.

6.Most  of the discussion is repetition of the result. Please delete and explore the result through analysis of the result 

In addition to some minor revisions.

Line 25, Please revise "COVID-19 is an infectious disease" as "Coronavirus Disease 2019 (COVID-19) is zoonotic infectious disease"

Please revise COVID-19 disease with "COVID-19" infection.

Line 8, delete "our," , because it globally effects the capacity protocol of hospitals.

On Line 9, add " AUC" to Intensive Care Units between brackets.

Lines 25-29, please add references. "Coronavirus disease 2019 (COVID-19) Situation Report – 57, WHO, 2020, 10 AM CET 17 March 2020"

**********

6. PLOS authors have the option to publish the peer review history of their article (what does this mean?). If published, this will include your full peer review and any attached files.

If you choose “no”, your identity will remain anonymous but your review may still be made public.

Do you want your identity to be public for this peer review? For information about this choice, including consent withdrawal, please see our Privacy Policy.

Reviewer #1: Yes: Alexandre de Fátima Cobre

Reviewer #2: No

**********

[NOTE: If reviewer comments were submitted as an attachment file, they will be attached to this email and accessible via the submission site. Please log into your account, locate the manuscript record, and check for the action link "View Attachments". If this link does not appear, there are no attachment files.]

While revising your submission, please upload your figure files to the Preflight Analysis and Conversion Engine (PACE) digital diagnostic tool, https://pacev2.apexcovantage.com/. PACE helps ensure that figures meet PLOS requirements. To use PACE, you must first register as a user. Registration is free. Then, login and navigate to the UPLOAD tab, where you will find detailed instructions on how to use the tool. If you encounter any issues or have any questions when using PACE, please email PLOS at figures@plos.org. Please note that Supporting Information files do not need this step.

PLoS One. 2023 May 19;18(5):e0283469. doi: 10.1371/journal.pone.0283469.r002

Author response to Decision Letter 0

31 Oct 2022

Dear Dr. Arteaga-Livias

Thanks for the opportunity to answer the reviewer’s comments. We are grateful for their carefully reading of our work and for their helpful advices and comments. We answer point-by-point each comment ad rewrite the manuscript incorporating these changes. We make an effort in explain better the algorithm and the used methodology an also the novelty and the focus of this work

REVIEWER 1:

Reviewer #1: Overall, the study was very well conducted. I start by congratulating the authors.

We thank the reviewer for their comment and their help to improve the manuscript

Below are my comments and I hope they are useful to improve your study.

1. Line 63, “Continuous variables have been described by the mean, minimum and maximum data”. Clinical data have almost always been shown not to have a normal distribution, my suggestion is that these results be presented in the form of median and interquartile range.

We totally agree with the reviewer´s suggestion, data presentation has been changed in the new manuscript version.

2. Line 73, “Statistical significance was defined as follows: p<0.001***”. Why are the levels of significance different? Was there any sampling procedure that was done to get to this point?

Our impression is that there is some misunderstanding. The type I level of significance has been set at 0.05, however in the charts, one star means that the p-value is below 0.05, two stars mean that is below 0.01, and three stars that is below 0.001. We will rewrite it in the new manuscript version.

3.On the lines 99-100, "The Model Training step is 100 described in Figure 1". I saw to figure 1, it's actually just a graph of the age distribution as a function of severity. I understood nothing. Are you sure it's figure 1?

Sorry about this mistake, as you notice there is an error in the order of the uploaded figures. Figure 1 is indeed a schematic description of the model.

4.Lines 100-101, “Finally, a Cross-Validation 102 method is applied to decrease overfitting impact”. What type of cross-validation was used? Venetian blind?, leave-one-out?

We use k-fold as a cross-validation method. This procedure has been used with a k =10 meaning a 10-fold cross validation. This information has been added to the new version of the manuscript.

5.I encourage the authors to insert the formulas for calculating the metrics for evaluating the performance of the models: sensitivity, specificity, accuracy (material and methods). Still on the evaluation of the performance of the models, I ask that the authors discuss the following scientific articles: PMID: 35751188, PMID: 34091385.

The formulas used are the ones by default in the used approach (SciPy):

- Sensitivity: True positives /(True positives+ False Negatives)

- Specificity: True Negative / (True negative + False Positives)

- accuracy: (True positive + True Negative) / (True Positive + True Negative + False Positive + False Negative)

We include in the discussion the proposed references.

6.Still on the evaluation of the performance of the models, I ask that authors present a table of confusion matrix with: true positives, true negatives, false positives, false negatives, sensitivity, specificity and accuracy. This must be shown for all classifiers used. For more understanding, see the PMID paper table: 35751188.

A new supplementary figure including the Table with the values for each model has been generated and added as Supplementary information.

7.What types of preprocessing are used for machine learning data? I encourage authors to clarify, in the material and methods section and in the Results section.

The dataset was constructed with the 42 features of 1082 patients, from these 37 variables were chosen avoiding redundancy. Aassociation and correlation between the variables were analyzed (AutoDiscovery Strength Score) to avoid model overfitting and to identify the patient's features that were more associated with the predictive goal. In this specific case, the association of the variables with the severity variable was evaluated and one of the correlated pair of variables was removed, if any. To do that, AutoDiscovery selected the proper numerical method based on the data type and distribution of the variables assessed. Numerical tests applied in each case are Spearman’s Rank Correlation, Variance Analysis (ANOVA one-way, U Mann-Whitney, and Kruskal-Wallis; normality was tested in these cases with D’Agostino/Pearson methods), and Cramer’s V Contingency Index. This massive data exploration process was carried out systematically for each patient group (stratum), considering only those correlations evaluated in a data subset with n≥5, two-tailed p < 0.05 and Spearman’s rank correlation coefficient > 0.8 to be relevant.

Given the nature of this multiple testing method, a high significance threshold was calculated based on Benjamini-Hochberg method (False-Discovery Rate) to classify the rest associations.

The Feature Selection process, where, based on Recursive Feature Elimination (RFE) and the AutoDiscovery Strength Score, age decade and the Lymphocytes to Immature Granulocyte ratio were added as variables for the model.

This information has been added to the manuscript.

 

REVIEWER 2:

The article needs revision and re-writing with good language and grammar.

Thanks to the reviewer for his/her comments that help us to improve the manuscript

1. What is the novelty of this research?

Although research in COVID19 moves very fast and the needs are changing, our work present6 an algorithm based in data that are obtained in all Hospitals routinely and no need for any extra analysis. That is novel and helpful for all the hospitals in the different countries to take fast decisions. Moreover our work underline the importance of the immature granulocites status in the gravity of the evolution.

2.What did the authors find new ?

We present an algorithm based in routinely parameters obtained in the clinical routine. Therefore decisions can be made in the first days of hospitalization without any additional analysis, just based in the regular data obtained the first day of hospitalization. We also highlight the importance of immature granulocytes in the evolution of the disease.

3. The data is old and dates back to the year 2020, and much research has been published on this topic.

We partially agree with the reviewer. We try to share our data as quickly as we could (and for that reason manuscript was located at BioRiv repository), however although the data are from the 2020, this is the most naïve data (without vaccinated people) and the ones that give us a more accurate idea about how the disease work. That could be useful for a new coronavirus pandemic and also to understand better the new cases that re arriving to the hospitals.

4. Why did the authors not focus on albumin levels, despite the fact that albumin was found to decrease in much of the research.

The aim of the work was not to select the biomarkers but to use the ones that are routinely produced in the regular clinical practice. In fact we recover the data from the clinical records with no interaction (at that moment) with the clinicians, meaning that the choice of variables was based in the clinical needs and in the regular protocols. Is an undirected approach with the aim to find an easy and cheap way to make decisions at the hospital.

5.Why did the authors not focus on LDH levels, despite the fact that albumin was found to increase in much of the research.

Please, see response to answer 4.

6.Most of the discussion is repetition of the result. Please delete and explore the result through analysis of the result

Thanks to the reviewer for his/her suggestions. We rewrite the discussion focusing in the analysis of the results.

Attachment

Submitted filename: response to reviewers_PLos One v 3.docx

Click here for additional data file. (241.5KB, docx)

Decision Letter 1

Kovy Arteaga-Livias

24 Nov 2022

PONE-D-22-16300R1Early biochemical analysis of COVID-19 patients helps severity predictionPLOS ONE

Dear Dr. Otaegui,

Thank you for submitting your manuscript to PLOS ONE. After careful consideration, we feel that it has merit but does not fully meet PLOS ONE’s publication criteria as it currently stands. Therefore, we invite you to submit a revised version of the manuscript that addresses the points raised during the review process.

Please submit your revised manuscript by Jan 08 2023 11:59PM. If you will need more time than this to complete your revisions, please reply to this message or contact the journal office at plosone@plos.org. When you're ready to submit your revision, log on to https://www.editorialmanager.com/pone/ and select the 'Submissions Needing Revision' folder to locate your manuscript file.

Please include the following items when submitting your revised manuscript:

  • A rebuttal letter that responds to each point raised by the academic editor and reviewer(s). You should upload this letter as a separate file labeled 'Response to Reviewers'.

  • A marked-up copy of your manuscript that highlights changes made to the original version. You should upload this as a separate file labeled 'Revised Manuscript with Track Changes'.

  • An unmarked version of your revised paper without tracked changes. You should upload this as a separate file labeled 'Manuscript'.

If you would like to make changes to your financial disclosure, please include your updated statement in your cover letter. Guidelines for resubmitting your figure files are available below the reviewer comments at the end of this letter.

If applicable, we recommend that you deposit your laboratory protocols in protocols.io to enhance the reproducibility of your results. Protocols.io assigns your protocol its own identifier (DOI) so that it can be cited independently in the future. For instructions see: https://journals.plos.org/plosone/s/submission-guidelines#loc-laboratory-protocols. Additionally, PLOS ONE offers an option for publishing peer-reviewed Lab Protocol articles, which describe protocols hosted on protocols.io. Read more information on sharing protocols at https://plos.org/protocols?utm_medium=editorial-email&utm_source=authorletters&utm_campaign=protocols.

We look forward to receiving your revised manuscript.

Kind regards,

Kovy Arteaga-Livias

Academic Editor

PLOS ONE

[Note: HTML markup is below. Please do not edit.]

Reviewers' comments:

Reviewer's Responses to Questions

Comments to the Author

1. If the authors have adequately addressed your comments raised in a previous round of review and you feel that this manuscript is now acceptable for publication, you may indicate that here to bypass the “Comments to the Author” section, enter your conflict of interest statement in the “Confidential to Editor” section, and submit your "Accept" recommendation.

Reviewer #1: (No Response)

Reviewer #2: All comments have been addressed

**********

2. Is the manuscript technically sound, and do the data support the conclusions?

The manuscript must describe a technically sound piece of scientific research with data that supports the conclusions. Experiments must have been conducted rigorously, with appropriate controls, replication, and sample sizes. The conclusions must be drawn appropriately based on the data presented.

Reviewer #1: Yes

Reviewer #2: Yes

**********

3. Has the statistical analysis been performed appropriately and rigorously?

Reviewer #1: No

Reviewer #2: Yes

**********

4. Have the authors made all data underlying the findings in their manuscript fully available?

The PLOS Data policy requires authors to make all data underlying the findings described in their manuscript fully available without restriction, with rare exception (please refer to the Data Availability Statement in the manuscript PDF file). The data should be provided as part of the manuscript or its supporting information, or deposited to a public repository. For example, in addition to summary statistics, the data points behind means, medians and variance measures should be available. If there are restrictions on publicly sharing data—e.g. participant privacy or use of data from a third party—those must be specified.

Reviewer #1: Yes

Reviewer #2: Yes

**********

5. Is the manuscript presented in an intelligible fashion and written in standard English?

PLOS ONE does not copyedit accepted manuscripts, so the language in submitted articles must be clear, correct, and unambiguous. Any typographical or grammatical errors should be corrected at revision, so please note any specific errors here.

Reviewer #1: Yes

Reviewer #2: Yes

**********

6. Review Comments to the Author

Please use the space provided to explain your answers to the questions above. You may also include additional comments for the author, including concerns about dual publication, research ethics, or publication ethics. (Please upload your review as an attachment if it exceeds 20,000 characters)

Reviewer #1: Dear Editor, I would like to thank you for the opportunity to review the article “Early biochemical analysis of COVID-19 patients helps severity prediction”, and I would like to take this opportunity to apologize for the delay in completing the review, I had unforeseen personal problems. Second, I would like to congratulate the authors for the article. Overall I enjoyed the post however some improvements need to be made. My observations follow.

1.In the “Abstract” section, the authors describe “…The test has been done by standardized cost-effective technique available in all the hospitals”. If the method is being used in clinical practice, it means that the model has been deployed. If so, I strongly recommend that you provide the access link for your Machine Learning App. I am just saying that the authors need to be realistic, because if the App has not been developed for the developed models, I recommend that they remove the phrase “…The test has been done by standardized cost-effective technique available in all the hospitals” in the manuscript.

2.In the material and methods section, “Continuous variables have been described by the mean, minimum and maximum data…”. I suggest initially testing the normality of the data. If normality is proven, the results must be presented as mean and standard deviation, otherwise the results must be presented as median and interquartile range.

3.The data used in this study are unbalanced: severe cases 449 (41.5%); ICU admission 60 (5.55%); deceased 140 (12.94%). It is important to point out that for imbalance problems, the AUC is not a better metric to evaluate the performance of the classifiers. The most reliable metric is the “F1 Score”. I strongly recommend that authors report these data in the article, including citing them in the abstract.

4.Still in material and methods, the authors describe “…Finally, a Cross-Validation 102 method is applied to decrease overfitting impact”. It needs to be clear which type of cross-validation was adopted.

5.The figures are very blurred. I ask you to improve the resolution of the figures. I also ask you to increase the size of the X and Y axes of the figures.

Reviewer #2: The author response to all reviewer comment , and the manuscript developed to the level of publication

**********

7. PLOS authors have the option to publish the peer review history of their article (what does this mean?). If published, this will include your full peer review and any attached files.

If you choose “no”, your identity will remain anonymous but your review may still be made public.

Do you want your identity to be public for this peer review? For information about this choice, including consent withdrawal, please see our Privacy Policy.

Reviewer #1: Yes: Alexandre de Fátima Cobre

Reviewer #2: No

**********

[NOTE: If reviewer comments were submitted as an attachment file, they will be attached to this email and accessible via the submission site. Please log into your account, locate the manuscript record, and check for the action link "View Attachments". If this link does not appear, there are no attachment files.]

While revising your submission, please upload your figure files to the Preflight Analysis and Conversion Engine (PACE) digital diagnostic tool, https://pacev2.apexcovantage.com/. PACE helps ensure that figures meet PLOS requirements. To use PACE, you must first register as a user. Registration is free. Then, login and navigate to the UPLOAD tab, where you will find detailed instructions on how to use the tool. If you encounter any issues or have any questions when using PACE, please email PLOS at figures@plos.org. Please note that Supporting Information files do not need this step.

PLoS One. 2023 May 19;18(5):e0283469. doi: 10.1371/journal.pone.0283469.r004

Author response to Decision Letter 1

12 Dec 2022

Dear Dr. Arteaga-Livias

Thanks for the opportunity to answer this second round of reviewer’s comments. We are grateful for their carefully reading of our work and for their helpful advices and comments. We answer point-by-point each comment ad rewrite the manuscript incorporating these changes.

REVIEWER 1:

Dear Editor, I would like to thank you for the opportunity to review the article “Early biochemical analysis of COVID-19 patients helps severity prediction”, and I would like to take this opportunity to apologize for the delay in completing the review, I had unforeseen personal problems. Second, I would like to congratulate the authors for the article. Overall I enjoyed the post however some improvements need to be made. My observations follow.

1.In the “Abstract” section, the authors describe “…The test has been done by standardized cost-effective technique available in all the hospitals”. If the method is being used in clinical practice, it means that the model has been deployed. If so, I strongly recommend that you provide the access link for your Machine Learning App. I am just saying that the authors need to be realistic, because if the App has not been developed for the developed models, I recommend that they remove the phrase “…The test has been done by standardized cost-effective technique available in all the hospitals” in the manuscript.

Sorry for the misunderstanding. The model has been developed from biochemical data obtained in a routine assay. This sentence wants to emphasize that this kind of data could be easily obtained in the vast majority of hospitals and therefore the methodology proposed in our work can be easily transferred to clinical practice. We change this phrase in the abstract to clarify the point.

2.In the material and methods section, “Continuous variables have been described by the mean, minimum and maximum data…”. I suggest initially testing the normality of the data. If normality is proven, the results must be presented as mean and standard deviation, otherwise the results must be presented as median and interquartile range.

Thanks for the suggestion. As requested by a reviewer we change the presentation of these data to mean, minimum and maximum because, according to the reviewer, is the way in which clinical data should be presented. Moreover we perform a Normality test (Shapiro- Wilk) to confirm that our the variables should be presented in this way.

3.The data used in this study are unbalanced: severe cases 449 (41.5%); ICU admission 60 (5.55%); deceased 140 (12.94%). It is important to point out that for imbalance problems, the AUC is not a better metric to evaluate the performance of the classifiers. The most reliable metric is the “F1 Score”. I strongly recommend that authors report these data in the article, including citing them in the abstract.

Thanks for the comment. In this work we do not design the groups but use the data from the “real world”, recovering the data from a real situation from our hospital during a period of time. Following your suggestion we add the F1-score value to the manuscript.

4.Still in material and methods, the authors describe “…Finally, a Cross-Validation 102 method is applied to decrease overfitting impact”. It needs to be clear which type of cross-validation was adopted.

We use k-fold as a cross-validation method. This procedure has been used with a k =10 meaning a 10-fold cross validation. This information has been added to the new version of the manuscript.

5.The figures are very blurred. I ask you to improve the resolution of the figures. I also ask you to increase the size of the X and Y axes of the figures.

Quality of the figures and Axes have been improved

REVIEWER 2:

The author response to all reviewer comment , and the manuscript developed to the level of publication

Attachment

Submitted filename: response to reviewers 2nd round.docx

Click here for additional data file. (14.2KB, docx)

Decision Letter 2

Kovy Arteaga-Livias

9 Mar 2023

Early biochemical analysis of COVID-19 patients helps severity prediction

PONE-D-22-16300R2

Dear Dr. Otaegui,

We’re pleased to inform you that your manuscript has been judged scientifically suitable for publication and will be formally accepted for publication once it meets all outstanding technical requirements.

Within one week, you’ll receive an e-mail detailing the required amendments. When these have been addressed, you’ll receive a formal acceptance letter and your manuscript will be scheduled for publication.

An invoice for payment will follow shortly after the formal acceptance. To ensure an efficient process, please log into Editorial Manager at http://www.editorialmanager.com/pone/, click the 'Update My Information' link at the top of the page, and double check that your user information is up-to-date. If you have any billing related questions, please contact our Author Billing department directly at authorbilling@plos.org.

If your institution or institutions have a press office, please notify them about your upcoming paper to help maximize its impact. If they’ll be preparing press materials, please inform our press team as soon as possible -- no later than 48 hours after receiving the formal acceptance. Your manuscript will remain under strict press embargo until 2 pm Eastern Time on the date of publication. For more information, please contact onepress@plos.org.

Kind regards,

Kovy Arteaga-Livias

Academic Editor

PLOS ONE

Additional Editor Comments (optional):

Reviewers' comments:

Reviewer's Responses to Questions

Comments to the Author

1. If the authors have adequately addressed your comments raised in a previous round of review and you feel that this manuscript is now acceptable for publication, you may indicate that here to bypass the “Comments to the Author” section, enter your conflict of interest statement in the “Confidential to Editor” section, and submit your "Accept" recommendation.

Reviewer #2: All comments have been addressed

**********

2. Is the manuscript technically sound, and do the data support the conclusions?

The manuscript must describe a technically sound piece of scientific research with data that supports the conclusions. Experiments must have been conducted rigorously, with appropriate controls, replication, and sample sizes. The conclusions must be drawn appropriately based on the data presented.

Reviewer #2: Yes

**********

3. Has the statistical analysis been performed appropriately and rigorously?

Reviewer #2: Yes

**********

4. Have the authors made all data underlying the findings in their manuscript fully available?

The PLOS Data policy requires authors to make all data underlying the findings described in their manuscript fully available without restriction, with rare exception (please refer to the Data Availability Statement in the manuscript PDF file). The data should be provided as part of the manuscript or its supporting information, or deposited to a public repository. For example, in addition to summary statistics, the data points behind means, medians and variance measures should be available. If there are restrictions on publicly sharing data—e.g. participant privacy or use of data from a third party—those must be specified.

Reviewer #2: Yes

**********

5. Is the manuscript presented in an intelligible fashion and written in standard English?

PLOS ONE does not copyedit accepted manuscripts, so the language in submitted articles must be clear, correct, and unambiguous. Any typographical or grammatical errors should be corrected at revision, so please note any specific errors here.

Reviewer #2: Yes

**********

6. Review Comments to the Author

Please use the space provided to explain your answers to the questions above. You may also include additional comments for the author, including concerns about dual publication, research ethics, or publication ethics. (Please upload your review as an attachment if it exceeds 20,000 characters)

Reviewer #2: All comments has been addressed .

The manuscript improved to the level of publication.......................

**********

7. PLOS authors have the option to publish the peer review history of their article (what does this mean?). If published, this will include your full peer review and any attached files.

If you choose “no”, your identity will remain anonymous but your review may still be made public.

Do you want your identity to be public for this peer review? For information about this choice, including consent withdrawal, please see our Privacy Policy.

Reviewer #2: No

**********

Acceptance letter

Kovy Arteaga-Livias

10 May 2023

PONE-D-22-16300R2

Early biochemical analysis of COVID-19 patients helps severity prediction

Dear Dr. Otaegui:

I'm pleased to inform you that your manuscript has been deemed suitable for publication in PLOS ONE. Congratulations! Your manuscript is now with our production department.

If your institution or institutions have a press office, please let them know about your upcoming paper now to help maximize its impact. If they'll be preparing press materials, please inform our press team within the next 48 hours. Your manuscript will remain under strict press embargo until 2 pm Eastern Time on the date of publication. For more information please contact onepress@plos.org.

If we can help with anything else, please email us at plosone@plos.org.

Thank you for submitting your work to PLOS ONE and supporting open access.

Kind regards,

PLOS ONE Editorial Office Staff

on behalf of

Dr. Kovy Arteaga-Livias

Academic Editor

PLOS ONE

Associated Data

    This section collects any data citations, data availability statements, or supplementary materials included in this article.

    Supplementary Materials

    S1 File. Supporting information, including one figure and raw data.

    (ZIP)

    Click here for additional data file. (180KB, zip)
    Attachment

    Submitted filename: response to reviewers_PLos One v 3.docx

    Click here for additional data file. (241.5KB, docx)
    Attachment

    Submitted filename: response to reviewers 2nd round.docx

    Click here for additional data file. (14.2KB, docx)

    Data Availability Statement

    All relevant data are within the paper and its Supporting Information files.

    Articles from PLOS ONE are provided here courtesy of PLOS

    RESOURCES