Bayesian Model Improves Antibody Titer Estimation for RSV Data

Researchers have developed a Bayesian hierarchical model (BHM) to enhance the accuracy and robustness of neutralizing antibody titer estimates, specifically applied to Respiratory Syncytial Virus (RSV) data. Published in Nature Communications, the study addresses significant batch-level biases and experimental noise inherent in traditional methods like the Kärber formula and four-parameter logistic (4PL) models. Through simulations and experimental data analysis, the BHM demonstrated superior performance, achieving a Spearman correlation of 0.96 with simulation truth, compared to 0.63 for the Kärber formula and 0.87 for the 4PL model. The new framework significantly reduced false negatives to 0.93%, whereas the Kärber formula yielded 9.85%. Furthermore, the study highlights that population-level metrics, such as geometric mean titers and seroprevalence, vary substantially depending on the estimation method used. This adaptable framework offers a more reliable tool for assessing population immunity and guiding vaccine development by correcting for experimental variations across different settings.