Antibody titers wane after two-dose COVID-19 vaccinations, but individual variation in vaccine-elicited antibody dynamics remains to be explored. Here, we created a personalized antibody score that enables individuals to infer their antibody status by use of a simple calculation. We recently developed a mathematical model of B cell differentiation to accurately interpolate the longitudinal data from a community-based cohort in Fukushima, Japan, which consists of 2,159 individuals who underwent serum sampling two or three times after a two-dose vaccination with either BNT162b2 or mRNA-1273. Using the individually reconstructed time course of the vaccine-elicited antibody response, we first elucidated individual background factors that contributed to the main features of antibody dynamics, i.e., the peak, the duration, and the area under the curve. We found that increasing age was a negative factor and a longer interval between the two doses was a positive factor for individual antibody level. We also found that the presence of underlying disease and the use of medication affected antibody levels negatively, whereas the presence of adverse reactions upon vaccination affected antibody levels positively. We then applied to these factors a recently proposed computational method to optimally fit clinical scores, which resulted in an integer-based score that can be used to evaluate the antibody status of individuals from their basic demographic and health information. This score can be easily calculated by individuals themselves or by medical practitioners. There is a potential usefulness of this score for identifying vulnerable populations and encouraging them to get booster vaccinations.Significance statementDifferent individuals show different antibody titers even after the same COVID-19 vaccinations, making some individuals more prone to breakthrough infections than others. Such variability remains to be clarified. Here we used mathematical modeling to reconstruct individual post-vaccination antibody dynamics from a cohort of 2,159 individuals in Fukushima, Japan. Machine learning identified several positive and negative factors affecting individual antibody titers. Positive factors included adverse reactions after vaccinations and a longer interval between two vaccinations. Negative factors included age, underlying medical conditions, and medications. We combined these factors and developed an “antibody score” to estimate individual antibody dynamics from basic demographic and health information. This score can help to guide individual decision-making about taking further precautions against COVID-19.