A cellular senescence-related gene signature for predicting severity in COVID-19 analyzed by bioinformatics combining machine learning algorithm

<bold>Abstract</bold> <bold>Background:</bold>The coronavirus disease 2019 (COVID-19) pandemic is recognized as a significant challenge to the public health and is associated with high morbidity and mortality. Cellular senescence plays an important role in the development of COVID-19. However, the potential prognostic value of cellular senescence-related gene signature in COVID-19 remains unknown. <bold>Objective:</bold>The study aimed to create a cellular senescence-related gene signature to evaluate the severity in COVID-19 patients and provide potential therapeutic targets. <bold>Methods:</bold>Consensus-weighted gene co-expression network analysis (WGCNA),functional enrichment and LASSO regression were used to construct a prognostic model.ESTIMATE,CIBERSORT and ssGSEA were performed to investigate the immune landscape in COVID-19 patients. <bold>Results:</bold>Six cellular senescence-related genes including CEBPB,CENPA,XAF1,MATK,UBTD1 and E2F1 were used to construct a predictive model to evaluate the severity in COVID-19 patients. The model had high prognostic accuracy for predicting severity in COVID-19 (admission to ICU and incidence of mechanical ventilation) and was correlation with clinical characteristics, such as HFD45,ventilator-free days,APACHE-II and SOFA. Innate and adaptive immune function and cells were significantly different between non-ICU group and ICU group. Six genes were significantly correlated with immune cells. <bold>Conclusions:</bold>In conclusion, a cellular senescence-related gene signature was established and validated for predicting severity in COVID-19,which may provide new insights into the roles of cellular senescence in COVID-19 and effective therapeutic approaches.