Which vaccine works best for you: Using pre-vaccination serology to provide more personalized vaccine recommendations 2109

Abstract Description Although influenza is one of the best-studied viruses, vaccine effectiveness remains around 20-50%. Surprisingly, a sizable fraction of individuals show little-to-no antibody response following vaccination, yet we cannot identify these individuals a priori nor ascertain whether a different vaccine would have served them better. This work will demonstrate how machine learning can use the wealth of prior studies to robustly forecast vaccine responses across 10+ influenza seasons. By leveraging model interpretability, we quantify the importance of factors such as age, sex, prior immunity, and the vaccine’s formulation, paving the way for more nuanced vaccine recommendations. A pre-vaccination feature - the time between peak HAI across recent variants - distinguished large versus small post-vaccination responses with 73% accuracy. This approach paves the way to better utilize current influenza vaccines, especially for individuals that exhibit the weakest responses. Funding Sources Supported by the University of California San Diego PREPARE Institute and by the Bodman family. Topic Categories Computational and Systems Immunology (COMP)