NIH Research Festival
Inference of nonlinear dynamics and parameters in biological modeling is a challenging task. Approaches relying on hypothetical underlying mechanisms can complicate the inference process because standard parameter optimization methods are difficult to constrain to physiological ranges. Is the model at fault or the parameter optimization? We propose an approach that utilizes neural networks to address parameter inference and physiological modeling simultaneously. In this study, we solve an optimization problem using a lipolysis model to obtain parameter values for a physiological model of the dynamics of glucose, insulin, and free fatty acids. First, we generate sample parameters, integrate the model with the sample parameters, and obtain simulated data. We then train a convolutional neural network to output the model parameters and evaluate its performance in reconstructing simulated or experimental data. Our objective is to use the power of deep learning methods to enhance modeling techniques and improve training processes by incorporating more nonlinear terms of state variables as inputs to the network.
Scientific Focus Area: Computational Biology
This page was last updated on Monday, September 25, 2023