Development and validation of a nomogram to predict severe influenza

Area of Science:

• Medical research
• Biostatistics
• Genomics

Background:

• Influenza is a severe acute respiratory illness with significant public health implications.
• Early identification of patients at risk for severe influenza can reduce mortality rates.

Purpose of the Study:

• To develop and validate a predictive model for severe influenza using clinical and molecular data.
• To assess the association between specific biomarkers and clinical factors with influenza severity.

Main Methods:

• Analysis of 146 influenza patients' data from the Gene Expression Omnibus (GEO) database.
• Utilized R software for variable selection, including Least Absolute Shrinkage and Selection Operator (LASSO) and multivariate logistic regression.
• Developed a nomogram for severe influenza prediction, validated using C-index, AUC, DCA, and calibration curves.

Main Results:

• Severe influenza was observed in 32.20% of patients and correlated with age and illness duration.
• Multivariate logistic regression identified myeloperoxidase (MPO) level, haptoglobin (HP) level, and duration of illness as significant predictors.
• The developed nomogram achieved a C-index and AUC of 0.904.

Conclusions:

• A nomogram incorporating MPO, HP levels, and illness duration effectively predicts severe influenza early.
• This model can aid in guiding prevention and treatment strategies for severe influenza cases.