Nowadays the information of future rates of systems produced in series appears to be crucial to the production plan, especially for remanufacturing departments. Several questions need to be answered: the number of cores to remanufacture the condition of those cores as well as the number of replacement system the market requires. Based on a bivariate prognosis model, a multivariate model is developed to take more variables of stress in field into account. What's more, the fitting process has been optimized by using neural networks. This paper presents all necessary tools to perform a multivariate lifetime prognosis as well as the implementation in MATLAB.