We are talking about future research directions, not what's possible today. Given the success AI has had in virtually every domain its been deployed in, there's no reason to think it won't do well simulating physiological processes.