We provide an overview of a new generation of AI models and the opportunities to develop interpretable population PK models in an accelerated approach. This ML approach integrates genetic algorithm, stochastic gates and NONMEM to develop interpretable population PK models. Genetic algorithm was applied for the structure model selection and nonlinear feature selection was implemented by introducing the stochastic gates to the input layer of a neural network. We evaluated this approach with multiple published datasets/models to demonstrate this approach can deliver comparable population PK models as developed by pharmacometricans, but with shorter delivery timelines. The models can be easily interpreted and can be used in interactions with regulatory agencies.
