Path losses are an important measure of the effectiveness of wireless communication systems exposed to the effects of radio propagation in the surrounding environment. For many years, the correlations proposed by various academics have been used to calculate path loss for waves propagating in various environments with constrained operational parameters. In this study, alternative model is presented, based on artificial neural network weights, to the log-normal shadowing model used in calculating path losses across concrete surfaces. With path loss as the target variable, the data of the physical separation between the wireless sensor node transmitters and receivers (d) and the radial angle of the receiver node position (θ) was fed into the neural network during its training phase. Network weights were then used to build a new PL prediction formula. This formula more correctly predicts the average PL in concrete surfaces as its results are compared with those of log-normal shadowing model, the FSPL model and the Two-Ray model across all full ranges of the experimental data yield mean absolute deviation values of 0.51%, 4.1%, 40.58%, and 28.79%, respectively.