Updating a Predictive Neural Network Controller in Simulink

I am trying to update a Predictive Neural Network Controller from the Deep Learning Toolbox.

I would like to periodically take the Predictive Controller block offline (using an Autotuned PID block to temporarily maintain the system), retrain the network, update the Predictive Controller block with the retrained network and put it back online.

I've gotten as far as updating a saved network in a MATLAB function, but I am lost as to how I would go about implementing said retrained network.

Is it possible to update a Predicitive Neural Network Controller block while Simulink is running?

[Abhay Sengar]

Can you please share the model, use case and the expected outcomes?