Integration of mathematical modelling and knowledge-based systems for simulations of biochemical processes

In this work, a hybrid modelling scheme that integrates available mathematical description with artificial intelligence techniques was developed to simulate complex biochemical processes. The hybrid model features the parallel arrangement of a numeric module and a knowledge-based module. The numeric module was composed of 44 process variables and 145 model parameters, which involves algorithmic calculation of the differential equations for the metabolic behaviour of the biochemical processes. The knowledge-based module was designed to carry out frame-based inference to determine the cause of the modelling discrepancy using the relationship between the derivation of the process variables and the model parameters. Two types of frames (e.g., 8 frames for the variable error information and 14 frames for the parameter adjustment) were built to support the decision-making process. This novel hybrid modelling method was evaluated by comparing the predictions of a hybrid model with conventional mathematical modelling. The result has shown that by using the proposed scheme, the discrepancy between the model predictions and the experimental data can be reduced, and that the predictive strength of the simulation can be enhanced.