Question 6
Domain 1: Agent Architecture, Design, and DevelopmentWhat architectural pattern would best support this systematic diagnostic reasoning?
Correct answer: B
Explanation
A logic tree with prompt chains fits systematic diagnostic reasoning because it breaks the task into ordered decision points, moving from "symptom categorization" to "disease category identification" and then "specific disease narrowing." Each node uses a specialized prompt and passes its output forward, which supports structured, stepwise inference and targeted test recommendations.
Why each option is right or wrong
A. Use a single comprehensive prompt that includes all diagnostic criteria and asks the LLM to analyze the case holistically and provide a diagnosis with reasoning.
B. Implement a logic tree with prompt chains where each node represents a decision point (symptom categorization → disease category identification → specific disease narrowing → test recommendations), with each stage using a specialized prompt and feeding results to the next stage.
The correct architecture is a staged decision tree because the task requires sequential narrowing: first classify the presenting symptoms, then map them to a disease class, then refine to a specific diagnosis before generating test suggestions. In exam terms, each node functions as a gated inference step, with the output of one prompt constraining the next, which is exactly what a prompt-chain logic tree is designed to do.
C. Fine-tune an LLM specifically on medical diagnostic cases so it internalizes the diagnostic process and can perform end-to-end diagnosis in a single inference.
D. Deploy multiple specialized medical agents (symptom analyzer, disease identifier, test recommender) that work independently and combine their outputs through voting.