We suggest a multi-process dynamic model and a sequential bayesian forecasting method of tumor-specific growth. The mixture model uses prior information obtained from the general population and becomes more individualized as more observations from the tumor are sequentially taken into account. In this study we propose utilizing all available tumor-specific information up to date to approximate the unknown multi-scale process of tumor growth over time, in a stochastic context. The validation of our approach was performed with experimental data from mice and the results show that after few observations from a tumor are obtained and included in the model, the latter becomes more individualized, in the sense that its parameters are adjusted in order to reect the growth of each individual tumor, yielding more precise estimates of its size.