Results: Although the environment contained relatively few samples compared to the number of features—making it prone to overfitting on specific training sets—PySR showed comparable generalization performance to conventional models. In contrast, PySR demonstrated a remarkable advantage in explainability. It exhibited a richer ability to search for equations than logistic regression, which is limited to simple linear forms. Unlike tree-based or boosting models that only provide explanations in terms of feature importance, symbolic regression provided explicit mathematical expressions for class boundaries, which is a distinct advantage in the medical domain where pattern identification and classification are critical. Conclusion: Symbolic regression, which represents individual doctors’ decision-making criteria in explicit and interpretable mathematical forms, holds two major implications. First, it enables doctors to objectively review their own clinical decision-making criteria, thereby providing clear feedback. Second, by integrating the equations derived from multiple doctors within the same school of Korean medicine, it allows the formulation of group-level equations, laying the groundwork for developing school-specific CDSS AI