Figure 3. Deep Learning Model Workflow.

Deep learning models are made using the training data set. Model parameters are refined and tuned until the error is minimized when making predictions in the training set. The model is then tested by making predictions on the test data set, which the model has not seen previously. Standard metrics for classification model evaluation include generation of a confusion matrix which breaks down where the misclassifications happened and a receiver operating characteristic (ROC) curve providing information on how model performance compares to random. This figure was created using BioRender (https://biorender.com/)