Multi-Level Model to Predict Antibody Response to Influenza Vaccine Using Gene Expression Interaction Network Feature...

Area of Science:

• Immunology and Computational Biology
• Genomics and Machine Learning Applications

Background:

• Vaccination is key for influenza prevention, but some individuals show reduced antibody response, increasing infection risk.
• Identifying individuals with poor vaccine response is vital for personalized health strategies.

Purpose of the Study:

• To develop a predictive model for influenza vaccine response using baseline gene expression and antibody titers.
• To identify individuals at higher risk of infection despite vaccination.

Main Methods:

• A multi-level machine learning strategy combining pre-vaccination antibody titers and gene expression network interactions.
• Clustering individuals by baseline antibody titers (HAI) to refine gene-based modeling.
• Utilizing a gene-association interaction network (GAIN) for feature selection and identifying interacting gene pairs.

Main Results:

• The multi-level model effectively predicts post-vaccination antibody response, particularly in individuals with high baseline titers.
• GAIN feature selection enhanced model generalizability and identified key genes in B Cell Receptor signaling and antigen processing.
• Stratifying by baseline HAI enabled more targeted gene-based predictive modeling.

Conclusions:

• A multi-level machine learning approach can accurately predict influenza vaccine response by integrating baseline antibody levels and gene expression data.
• The developed model and interactive tool offer a promising strategy for identifying at-risk individuals and could be adapted for other vaccine studies.