A neural-network-based approach to rock-physics modeling in the context of time-lapse seismic analysis is presented. The aim is to find the optimum relationship between pressure, saturation, and seismic velocity. Such a relationship is essential in time-lapse interpretation, especially with reservoir-parameter inversion and forward modeling. The approach creates an opportunity to study reservoir behavior at the seismic level. Ultimately, time-lapse seismic interpretation can be improved if a better understanding of the relationship between rock and seismic properties is obtained.