Extracting relevant predictive variables for COVID-19 severity prognosis: An exhaustive comparison of feature selection techniques

Abstract

With the COVID-19 pandemic having caused unprecedented numbers of infections and deaths, large research efforts have been undertaken to increase our understanding of the disease and the factors which determine diverse clinical evolutions. Here we focused on a fully data-driven exploration regarding which factors (clinical or otherwise) were most informative for SARS-CoV-2 pneumonia severity prediction via machine learning (ML). In particular, feature selection techniques (FS), designed to reduce the dimensionality of data, allowed us to characterize which of our variables were the most useful for ML prognosis. We conducted a multi-centre clinical study, enrolling n = 1548 patients hospitalized due to SARS-CoV-2 pneumonia: where 792, 238, and 598 patients experienced low, medium and high-severity evolutions, respectively. Up to 106 patient-specific clinical variables were collected at admission, although 14 of them had to be discarded for containing ⩾60% missing values. Alongside 7 socioeconomic attributes and 32 exposures to air pollution (chronic and acute), these became d = 148 features after variable encoding. We addressed this ordinal classification problem both as a ML classification and regression task. Two imputation techniques for missing data were explored, along with a total of 166 unique FS algorithm configurations: 46 filters, 100 wrappers and 20 embeddeds. Of these, 21 setups achieved satisfactory bootstrap stability (⩾0.70) with reasonable computation times: 16 filters, 2 wrappers, and 3 embeddeds. The subsets of features selected by each technique showed modest Jaccard similarities across them. However, they consistently pointed out the importance of certain explanatory variables. Namely: patient’s C-reactive protein (CRP), pneumonia severity index (PSI), respiratory rate (RR) and oxygen levels –saturation Sp O2, quotients Sp O2/RR and arterial Sat O2/Fi O2–, the neutrophil-to-lymphocyte ratio (NLR) –to certain extent, also neutrophil and lymphocyte counts separately–, lactate dehydrogenase (LDH), and procalcitonin (PCT) levels in blood. A remarkable agreement has been found a posteriori between our strategy and independent clinical research works investigating risk factors for COVID-19 severity. Hence, these findings stress the suitability of this type of fully data-driven approaches for knowledge extraction, as a complementary to clinical perspectives.

Identification of Biomarkers for Severity in COVID-19 Through Comparative Analysis of Five Machine Learning Algoritms

“…The genetic algorithm has proven effective in several contexts, including the work of Kabir et al, 20 which introduced a redundancy reduction approach. In our study and in that of Hayet et al, 29 the genetic algorithms provided remarkable results, although the choice of appropriate objective functions could have improved the performance. Focusing on the detection of COVID-19, the study by Hayet et al, 29 highlights the importance of specific variables, such as CRP, Respiratory Rate, Oxygen Saturation, and LDH.…”

“…In this work, we sought to use diverse algorithms inspired by evolution, physical mechanisms and collective intelligence, as mentioned in the work of Agrawa et al 28 It should be noted that the algorithms applied were quite simple, trying to preserve the minimum version of each one to encourage equality of conditions, so they cannot be directly comparable with those of other works that present cumulative improvements said algorithms. 20,29,21,22,23 Relevant differences with said works will be discussed below.…”

Obtaining patient phenotypes in SARS-CoV-2 pneumonia, and their association with clinical severity and mortality

“…Table 3 covers the main clinical characterization of our clustering results per explanatory variable; focused on those 18 / 92 variables with statistically large inter-phenotype effect sizes. This subset of 18 key variables can also be viewed as a data-driven selection of the most informative factors for predicting the clinical outcomes under study (severity, mortality) [24,29]. In particular, phenotype C (low prevalence, 9.0%) included older patients, with more comorbidities, worse respiratory status (peripheral oxygenation, as well as in the arterial blood gas tests), and more unfavourable inflammatory, renal and/or hematologic biomarkers (C-reactive protein, procalcitonin, D-dimer, neutrophils-to-lymphocyte ratio, creatinine, BUN, prothrombin, etc.)…”

“…From the demographic and clinical information collected at the baseline time of hospitalization, 92 explanatory variables met our criterion of <60% missingness: whereas other 14 variables -e.g. ferritin, bilirubin, albumin, troponin, interleukin-6 (IL-6), aspartate aminotransferase (AST), creatine phosphokinase (CPK), platelets or eosinophils-failed to match this data quality criterion (see [24] for further details). Once categorical variables were transformed via one-hot encoding, these 92 attributes became d=109 features.…”

Feature identification using hypotheses of relevance and a 2D-cascade of SEQENS ensembles

“…This process removes irrelevant and redundant features, while the most discriminative ones are kept, allowing for training predictive models with higher discrimination power and better performance. Feature selection is evaluated by measuring the quality of the classification obtained by the selected subset of features (Hayet-Otero et al, 2023;Dabba et al, 2021;Bommert et al, 2020;Liu et al, 2002). Similarly to feature importance, feature selection is also a model-dependent approach.…”