Graduate Student University of Connecticut Storrs Mansfield, Connecticut
Liposome processing in a continuous manner can be really advantageous in terms of process reliability and scale up issues. However, controlling the critical quality attributes (CQAs) such as the particle size and polydispersity index can be challenging when different input parameters are involved. In such a case, an artificial neural network can be used as a predictive model to predict the CQAs of the liposome. This will help to cut down the experiments by getting an estimate of the process parameters to be used for the particular CQA. A feedforward, backpropagation neural network with multiple hidden layers can help in reducing the burden of experiments. Different models such as multiple-input multiple-output (MIMO) and multiple-input single-output (MISO) can be tried to get the best model. Incorporating molecular descriptors to give more information about the lipids can help to train the neural network.
Understand the basics of neural networks.
Understand the importance and utility of using predictive models.
Knowledge of different types of models and molecular descriptors.