Artificial neural network modelling has been preliminarily employed to investigate effects on the biodynamic responses. In order to evaluate the vibration transmission characteristics of the seat‑occupant system, further quantitative research is needed. Drawing from a low frequency experimental investigation into whole body vibration, this study is aimed to develop an ANN model with the response surface method optimization. The age, stature, sitting height, knee height, buttock‑to‑knee, weight, gender, BMI, cushion thickness and frequency are used as network input to explore that these how to predict transmissibility from the seat base to the seat pan. Based on the interaction between hyperparameters, the mapping relationship between model hyperparameters and prediction performance indexes was established, and the optimal combination of hyperparameters was optimized and obtained. The results show that the resonance frequencies in the vertical inline and the fore‑and‑aft cross‑axis transmissibilities from seat base to seat pan decreased with increasing thickness of foam at the seat pan. BP‑ANN model has good performance in establishing the nonlinear relationship between the anthropometric, seat structure characteristics and vibration transmission characteristics of seat‑occupant system. Compared with BP‑ANN model, the error of RSM‑BP‑ANN model is reduced by 25% and 18% respectively in predicting vertical in‑line transmissibility and fore‑and‑aft cross‑axis transmissibility from seat base to seat pan. And this also provides an idea for adjusting the parameters of neural network models to improve the prediction accuracy of seat transmissibility.