Development of Predictive Models for Disease Progression and Outcomes in Severe COVID-19 Patients Caused by Omicron Variants Using Metabolomics and Machine Learning Techniques

The ongoing pandemic of COVID-19 has entered a new phase. Despite the reduced pathogenicity of the currently prevalent Omicron variant of SARS-CoV-2, there is still a risk of severe illness and death. Currently, there are no specific biomarkers available to accurately predict the progression and outcomes of Omicron induced COVID-19. Previous studies have mainly focused on untargeted or lipid metabolism analysis of individuals infected with the original strain of SARS-CoV-2 or Omicron with mild to moderate symptoms. Therefore, we conducted a comprehensive targeted serum metabolomics analysis using wide-targeted metabolomics technology to analyze the metabolic profiles of COVID-19 patients infected with Omicron. We also correlated differentially expressed metabolites with laboratory test parameters. Finally, we developed a machine learning model that can accurately predict key biomarkers for the progression and prognosis of severe COVID-19, aiming to provide valuable evidence for improving the prognosis and reducing mortality in severe cases caused by Omicron.