Propane-air detonation is used as an example to illustrate a generalized and accurate method to predict the induction phase, heat release profile, and equilibrium state in complex hydrocarbon reactions. This is accomplished by representing pre-integrated heat-release and species information during the reaction via a mapping methodology, external to the fluid dynamic solver. The mapping, nondimensionalized and scaled in chemical time, uses a simplified system-entropy parameter to track reaction progress. Multilayer feedforward neural networks are used as function approximators to reproduce the parameters extracted from the detailed integrations and to perform the non-linear interpolations required between reaction points. Several sample cases for different reaction initial conditions are evaluated and presented. The mapping method results are accurate to within 1% to 3% to the results of detailed integrations, and the computational effort is reduced by two orders of magnitude. In problems of practical interest, further reduction in computational times are achievable, because this mapping method removes the numerical stiffness of the chemical reaction equations from the fluid dynamic solver and reduces the number of conservation equations. This approach is shown to work satisfactorily for rapidly compressed, non-diffusing, premixed near-stoichiometric reactions of propane and air under constant volume and constant pressure assumptions, forming the conceptual framework for potential extensions of the method to general reaction and combustion processes.