Predicting Alzheimer's disease based on genetic aspects and lifestyle plays a major role in healthcare field. In this paper, we present a software solution that constructs a statistical model based on patient dataset. The outcome model allows predicting efficiently the Alzheimer's disease for eventual patients. The multiple regression is used in the current paper to construct the statistical model. As well-known regression method involves the least squares estimation problem to estimate the parameters. The least squares estimation problem is solved by means of adjoint method. The computational time taken by algebraic adjoint method is colossal. Thus, the adjoint method is parallelized for the first time via MapReduce. The experimental results confirm the robustness of the outcome statistical model. Furthermore, the experiments show that the computational time is impressively reduced by the use of MapReduce.