In this paper, a brain tissue differential pathlength factor prediction model based on GS-SVM is established to quickly predict the DPF value of a brain tissue, and this method is compared with the BP-ANN prediction model. The results show that the grid optimization algorithm can automatically and accurately optimize the penalty parameter C and the parameter g of the Gaussian kernel function. The prediction results of the brain tissue differential pathlength factor prediction model based on GS-SVM are better than those based on BP-ANN, and have significant correlation with the prediction results of the Monte Carlo simulation method. It is expected to replace the Monte Carlo simulation method for batch calculation of DPF values. It can be applied in the near infrared cerebral oxygen monitoring instrument to make the calculation of physiological parameters of cerebral oxygen metabolism more rapid and accurate.