The importance of trust, consistency, and explainability for AI in health care

Think of an AI application in health care, real or hypothetical. Consider the criteria of trust, consistency, and explainability in relation to your chosen example. Consider the following points when sharing your ideas with your peers:

Trust: While the model performs well, on average, for the test set, can you trust that the model will work well given the specific patient you are serving?

Consistency: Is your model consistent with what you already know about the disease or problem?

Explainability: What has the model learned? What relationships have your model inferred from the training data?