A large-scale machine learning study of sociodemographic factors contributing to COVID-19 severity.

Understanding sociodemographic factors behind COVID-19 severity relates to significant methodological difficulties, such as differences in testing policies and epidemics phase, as well as a large number of predictors that can potentially contribute to severity. To account for these difficulties, we assemble 115 predictors for more than 3,000 US counties and employ a well-defined COVID-19 severity measure derived from epidemiological dynamics modeling. We then use a number of advanced feature selection techniques from machine learning to determine which of these predictors significantly impact the disease severity. We obtain a surprisingly simple result, where only two variables are clearly and robustly selected-population density and proportion of African Americans. Possible causes behind this result are discussed. We argue that the approach may be useful whenever significant determinants of disease progression over diverse geographic regions should be selected from a large number of potentially important factors.

Vitamin D, Albumin, and D-Dimer as Significant Prognostic Markers in Early Hospitalization in Patients with COVID-19.

SARS-CoV-2 continues to pose a major challenge to scientists and clinicians. We examined the significance of the serum concentrations of vitamin D, albumin, and D-dimer for the severity of the clinical picture and mortality in COVID-19. MATERIALS AND METHODS: A total of 288 patients treated for COVID-19 infection participated in the research. The patients were treated in the period from May 2020 to January 2021. All patients were divided based on the need for oxygen therapy (Sat > 94%) into patients with mild or severe clinical pictures. The biochemical and radiographic parameters of the patients were analyzed. Appropriate statistical methods were used in the statistical analysis. RESULTS: In patients with COVID-19 with confirmed severe clinical pictures, lower values of serum albumin (p < 0.0005) and vitamin D (p = 0.004) were recorded, as opposed to elevated values of D-dimer (p < 0.0005). Accordingly, the patients with fatal disease outcomes had lower levels of albumin (p < 0.0005) and vitamin D (p = 0.002), while their D-dimer (p < 0.0005) levels were elevated. An increase in the radiographic score, as a parameter for assessing the severity of the clinical picture, was accompanied by a decrease in serum albumin (p < 0.0005) and a simultaneous increase in D-dimer (p < 0.0005), without a change in the vitamin D concentration (p = 0.261). We also demonstrated the interrelations of the serum levels of vitamin D, albumin, and D-dimer in patients with COVID-19 as well as their significance as predictors of the outcome of the disease. CONCLUSION: The significance of the predictive parameters in our study indicates the existence of an important combined role of vitamin D, albumin, and D-dimer in the early diagnosis of the most severe patients suffering from COVID-19. Reduced values of vitamin D and albumin, in combination with elevated values of D-dimer, can be timely indicators of the development of a severe clinical picture and death due to COVID-19.

Galectin-3 as an important prognostic marker for COVID-19 severity.

Galectin-3 (Gal-3), multifunctional protein plays important roles in inflammatory response, infection and fibrosis. The goal of study was to determine the association of Gal-3, immune response, clinical, biochemical, and radiographic findings with COVID-19 severity. Study included 280 COVID-19 patients classified according to disease severity into mild, moderate, severe and critical group. Cytokines, clinical, biochemical, radiographic data and peripheral blood immune cell make up were analyzed. Patients in critical group had significantly higher serum level of Gal-3, IL-1ß, TNF-α, IL-12, IL-10 compared to the patients in less severe stages of disease. Strong positive correlation was detected between Gal-3 and IL-1ß, moderate positive correlation between Gal-3, TNF-α and IL-12, moderate negative correlation between Gal-3, IL-10/IL-1ß and IL-10/TNF-α. Moderate positive correlation noted between Gal-3 and urea, D dimer, CXR findings. Strong negative correlation detected between Gal-3 and p02, Sa02, and moderate negative correlation between Gal-3, lymphocyte and monocyte percentage. In the peripheral blood of patients with more severe stages of COVID-19 we detected significantly increased percentages of CD56- CD3+TNF-α+T cells and CD56- CD3+Gal-3+T cells and increased expression of CCR5 in PBMCs. Our results predict Gal-3 as an important marker for critical stage of COVID-19. Higher expression of Gal-3, TNF-α and CCR5 on T cells implicate on promoting inflammation and more severe form of disease.

Analyzing the GHSI puzzle of whether highly developed countries fared worse in COVID-19.

Global Health Security Index (GHSI) categories are formulated to assess the capacity of world countries to deal with infectious disease risks. Thus, higher values of these indices were expected to translate to lower COVID-19 severity. However, it turned out to be the opposite, surprisingly suggesting that higher estimated country preparedness to epidemics may lead to higher disease mortality. To address this puzzle, we: (i) use a model-derived measure of COVID-19 severity; (ii) employ a range of statistical learning approaches, including non-parametric machine learning methods; (iii) consider the overall excess mortality, in addition to official COVID-19 fatality counts. Our results suggest that the puzzle is, to a large extent, an artifact of oversimplified data analysis and a consequence of misclassified COVID-19 deaths, combined with the higher median age of the population and earlier epidemics onset in countries with high GHSI scores.

Understanding risk factors of a new variant outburst through global analysis of Omicron transmissibility.

The emergence of a new virus variant is generally recognized by its usually sudden and rapid spread (outburst) in a certain world region. Due to the near-exponential rate of initial expansion, the new strain may not be detected at its true geographical origin but in the area with the most favorable conditions leading to the fastest exponential growth. Therefore, it is crucial to understand better the factors that promote such outbursts, which we address in the example of analyzing global Omicron transmissibility during its global emergence/outburst in November 2021-February 2022. As predictors, we assemble a number of potentially relevant factors: vaccinations (both full and boosters), different measures of population mobility (provided by Google), estimated stringency of measures, the prevalence of chronic diseases, population age, the timing of the outburst, and several other socio-demographic variables. As a proxy for natural immunity (prevalence of prior infections in population), we use cumulative numbers of COVID-19 deaths. As a response variable (transmissibility measure), we use the estimated effective reproduction number (Re) averaged in the vicinity of the outburst maxima. To select significant predictors of Re, we use machine learning regressions that employ feature selection, including methods based on ensembles of decision trees (Random Forest and Gradient Boosting). We identify the young population, earlier infection onset, higher mobility, low natural immunity, and low booster prevalence as likely direct risk factors. Interestingly, we find that all these risk factors were significantly higher for Africa, though curiously somewhat lower in Southern African countries (where the outburst emerged) compared to other African countries. Therefore, while the risk factors related to the virus transmissibility clearly promote the outburst of a new virus variant, specific regions/countries where the outburst actually happens may be related to less evident factors, possibly random in nature.

Galectin-1 as the new player in staging and prognosis of COVID-19.

A new virus from the group of coronaviruses was identified as the cause of atypical pneumonia and called Severe Acute Respiratory Syndrome Coronavirus-2 (SARS-CoV-2) and disease called Corona Virus Disease (COVID-19). During the cytokine storm, the main cause of the death, proinflammatory cytokines are released which stimulate further tissue destruction. Galectin-1 (Gal-1) is a pleiotropic cytokine involved in many immune and inflammatory processes and its role in COVID-19 is still unknown. The aim of this study was to determine systemic values of Gal-1 and correlations between Gal-1 and proinflammatory cytokines and clinical parameters during COVID-19 progression. This is observational and cross-sectional study. 210 COVID-19 patients were included and divided into mild, severe or critical group according to COVID-19 severity. Serum levels of IL-1ß, IL-6, IL-10, IL-23, IL-33 and Gal-1 were measured using sensitive enzyme-linked immunosorbent assay (ELISA) kits. Systemic levels of IL-1ß, IL-6, IL-10, IL-23, IL-33 and Gal-1 were significantly higher in stage III of COVID-19 patients compared to stage I and II. There were no significant differences in the ratio between Gal-1 and IL-10 with proinflammatory cytokines. Positive correlation was detected between Gal-1 and IL-1ß, IL6, IL-10, IL-23 and IL-33. Gal-1 positively correlated with chest radiographic finding, dry cough and headache and negatively correlated with normal breathing sound. Linear regression model and ROC curve analysis point on Gal-1 as significant predictor for COVID-19 severity. Presented results implicate on Gal-1 and IL-10 dependent immunomodulation. The precise mechanism of Gal-1 effect in COVID-19 and its potential as a stage marker of disease severity is still to be clarified.

IL 33 Correlates With COVID-19 Severity, Radiographic and Clinical Finding.

Objective: The increased level of interleukin (IL)-33 is considered as a predictor of severe coronavirus disease 2019 (COVID-19) infection, but its role at different stages of the disease is still unclear. Our goal was to analyze the correlation of IL-33 and other innate immunity cytokines with disease severity. Methods: In this study, 220 patients with COVID-19 were included and divided into two groups, mild/moderate and severe/critical. The value of the cytokines, clinical, biochemical, radiographic data was collected and their correlation with disease severity was analyzed. Results: Most patients in the severe/critical group were male (81.8%) and older (over 64.5 years). We found a statistically significant difference (p < 0.05) in these two groups between clinical features (dyspnea, dry cough, fatigue, and auscultatory findings); laboratory [(neutrophil count, lymphocyte count, monocyte count, hemoglobin, plasma glucose, urea, creatinine, total bilirubin (TBIL), direct bilirubin (DBIL), aspartate aminotransferase (AST), albumin (ALB), lactate dehydrogenase (LDH), creatinine kinase (CK), D-dimer, C-reactive protein (CRP), procalcitonin (PCT), Fe, and Ferritin)], arterial blood gases (oxygen saturation-Sa02, partial pressure of oxygen -p02), and chest X-rays (CXR) lung findings (p = 0.000). We found a significantly higher serum concentration (p < 0.05) of TNF-α, IL-1ß, IL-6, IL-12, IL-23, and IL-33 in patients with COVID-19 with severe disease. In the milder stage of COVID-19, a positive correlation was detected between IL-33 and IL-1ß, IL-12 and IL-23, while a stronger positive correlation between the serum values of IL-33 and TNF-α, IL-1ß, IL-6, and IL-12 and IL-23 was detected in patients with COVID-19 with severe disease. A weak negative correlation (p < 0.05) between pO2 and serum IL-1ß, IL-12, and IL-33 and between SaO2 and serum IL-33 was noted. The positive relation (p < 0.05) between the serum values of IL-33 and IL-12, IL-33 and IL-6, and IL-6 and IL-12 is proven. Conclusion: In a more progressive stage of COVID-19, increased IL-33 facilitates lung inflammation by inducing the production of various innate proinflammatory cytokines (IL-1ß, IL-6, TNF-α, IL-12, and IL-23) in several target cells leading to the most severe forms of the disease. IL-33 correlates with clinical parameters of COVID-19 and might represent a promising marker as well as a therapeutic target in COVID-19.

COVID-19 severity determinants inferred through ecological and epidemiological modeling.

Understanding variations in the severity of infectious diseases is essential for planning proper mitigation strategies. Determinants of COVID-19 clinical severity are commonly assessed by transverse or longitudinal studies of the fatality counts. However, the fatality counts depend both on disease clinical severity and transmissibility, as more infected also lead to more deaths. Instead, we use epidemiological modeling to propose a disease severity measure that accounts for the underlying disease dynamics. The measure corresponds to the ratio of population-averaged mortality and recovery rates (m/r), is independent of the disease transmission dynamics (i.e., the basic reproduction number), and has a direct mechanistic interpretation. We use this measure to assess demographic, medical, meteorological, and environmental factors associated with the disease severity. For this, we employ an ecological regression study design and analyze different US states during the first disease outbreak. Principal Component Analysis, followed by univariate, and multivariate analyses based on machine learning techniques, is used for selecting important predictors. The usefulness of the introduced severity measure and the validity of the approach are confirmed by the fact that, without using prior knowledge from clinical studies, we recover the main significant predictors known to influence disease severity, in particular age, chronic diseases, and racial factors. Additionally, we identify long-term pollution exposure and population density as not widely recognized (though for the pollution previously hypothesized) significant predictors. The proposed measure is applicable for inferring severity determinants not only of COVID-19 but also of other infectious diseases, and the obtained results may aid a better understanding of the present and future epidemics. Our holistic, systematic investigation of disease severity at the human-environment intersection by epidemiological dynamical modeling and machine learning ecological regressions is aligned with the One Health approach. The obtained results emphasize a syndemic nature of COVID-19 risks.

Inferring the Main Drivers of SARS-CoV-2 Global Transmissibility by Feature Selection Methods.

Identifying the main environmental drivers of SARS-CoV-2 transmissibility in the population is crucial for understanding current and potential future outbursts of COVID-19 and other infectious diseases. To address this problem, we concentrate on the basic reproduction number R 0, which is not sensitive to testing coverage and represents transmissibility in an absence of social distancing and in a completely susceptible population. While many variables may potentially influence R 0, a high correlation between these variables may obscure the result interpretation. Consequently, we combine Principal Component Analysis with feature selection methods from several regression-based approaches to identify the main demographic and meteorological drivers behind R 0. We robustly obtain that country's wealth/development (GDP per capita or Human Development Index) is the most important R 0 predictor at the global level, probably being a good proxy for the overall contact frequency in a population. This main effect is modulated by built-up area per capita (crowdedness in indoor space), onset of infection (likely related to increased awareness of infection risks), net migration, unhealthy living lifestyle/conditions including pollution, seasonality, and possibly BCG vaccination prevalence. Also, we argue that several variables that significantly correlate with transmissibility do not directly influence R 0 or affect it differently than suggested by naïve analysis.

PM2.5 as a major predictor of COVID-19 basic reproduction number in the USA.

Many studies have proposed a relationship between COVID-19 transmissibility and ambient pollution levels. However, a major limitation in establishing such associations is to adequately account for complex disease dynamics, influenced by e.g. significant differences in control measures and testing policies. Another difficulty is appropriately controlling the effects of other potentially important factors, due to both their mutual correlations and a limited dataset. To overcome these difficulties, we will here use the basic reproduction number (R0) that we estimate for USA states using non-linear dynamics methods. To account for a large number of predictors (many of which are mutually strongly correlated), combined with a limited dataset, we employ machine-learning methods. Specifically, to reduce dimensionality without complicating the variable interpretation, we employ Principal Component Analysis on subsets of mutually related (and correlated) predictors. Methods that allow feature (predictor) selection, and ranking their importance, are then used, including both linear regressions with regularization and feature selection (Lasso and Elastic Net) and non-parametric methods based on ensembles of weak-learners (Random Forest and Gradient Boost). Through these substantially different approaches, we robustly obtain that PM2.5 is a major predictor of R0 in USA states, with corrections from factors such as other pollutants, prosperity measures, population density, chronic disease levels, and possibly racial composition. As a rough magnitude estimate, we obtain that a relative change in R0, with variations in pollution levels observed in the USA, is typically ~30%, which further underscores the importance of pollution in COVID-19 transmissibility.