You have now encountered two kinds of diagnostic system. The first, the animal classifier, was written by converting each rule directly into a Prolog clause, then relying on Prolog's question-answering ability to get results.
The second, the depression system, worked at a different level. Instead of translating each rule directly to Prolog, I followed the Winter Aid Fund quote, and built a kind of network. Think in terms of rooms, where each room contains a question. You take a different exit for each possible answer to this question, thus coming to a new room in which there sits either another question or a diagnosis. You may find this way of doing things useful for your project: many students do.