A Comprehensive Machine Learning Approach for COVID-19 Target Discovery in the Small-Molecule Metabolome

Abstract

Abstract Respiratory viruses, including influenza, RSV, and COVID-19, cause various respiratory infections. Distinguishing these viruses relies on diagnostic methods such as PCR testing. Challenges stem from overlapping symptoms and the emergence of new strains. Advanced diagnostics are crucial for accurate detection and effective management. This study leveraged nasopharyngeal metabolome data to predict respiratory virus scenarios including control vs RSV, control vs influenza A, control vs COVID-19, control vs all respiratory viruses, and COVID-19 vs influenza A/RSV. Our advanced machine learning models, including linear discriminant analysis, support vector machine, random forest, and logistic regression, exhibited superior accuracy, sensitivity, and specificity to previous supervised machine learning approaches. Key techniques such as feature ranking, standard scaling, and SMOTE were used to address class imbalances, thus enhancing model robustness. SHAP analysis identified crucial metabolites influencing positive predictions, thereby providing valuable insights into diagnostic markers. Our approach not only outperformed existing methods but also revealed top dominant features for predicting COVID-19, including Lysophosphatidylcholine acyl C18:2, Kynurenine, Phenylalanine, Valine, tyrosine, and aspartic Acid (Asp). These compounds play critical roles in metabolic pathways and have been identified as top contributors to predictive models in COVID-19 respiratory virus scenarios.