Guided Discovery for Clinical Reasoning Stability: A Lightweight Framework for Reducing Diagnostic Drift

This conceptual preprint introduces Guided Discovery (GD), a lightweight cognitive‑structural mechanism designed to detect and stabilize collapse‑vector patterns in clinical reasoning. GD reduces diagnostic drift by identifying instability in reasoning trajectories and re‑anchoring decision‑making to coherent diagnostic features. The paper presents two illustrative demos: one focused on human clinical reasoning and one on AI‑assisted triage. Together, they demonstrate how GD improves diagnostic consistency without requiring empirical datasets or algorithmic modification. GD offers a domain‑agnostic cognitive layer for clinicians, educators, and AI developers, and provides a foundation for future hybrid human‑AI reasoning systems.