This paper uses Bayesian networks for the state combination of random variables to describe the relationship between prior probability and posterior probability by likelihood. Meanwhile, the elimination of complex variables is achieved by combining the operation results of multivariate non-independent joint conditional probability decomposition. Based on this, Bayes’ theorem is used to calculate the learning efficiency, the comprehensive pass rate of any interactive courseware, and the growing value of the visualized learning process to improve the sharing of resources for senior education. To verify the feasibility of the model, a teaching effectiveness test was conducted, and the results showed that 66.6% of older adults believed that participation in senior education could strengthen their bodies and minds. The percentage of medical and health care knowledge courses was 63.20%. The direct effect of the learning environment on the participation behavior of senior education was 0.143, while the indirect effect of educational resources on the participation behavior of senior education was 0.083. In addition, 61.78% of seniors thought that attitude towards senior education was very important. Therefore, it can be concluded that the model of the elderly education system fills the gap in the structure of the lifelong education system, enriches the uppermost structure of the lifelong education system, and promotes the improvement and development of the lifelong education system.