From parallel to aligned: a United States Medical Licensing Examination (USMLE®) mapping and assessment-tagging framework for medical education

Abstract Background Medical students often experience a “parallel curriculum,” which is perceived or a real misalignment between what is taught in the formal institutional curriculum and, for licensing, the United States Medical Licensing Examination (USMLE®) content that they must master independently. By delivering transparency to the faculty and students, curriculum mapping institutional curriculum objectives and assessments to USMLE® content can reduce this perception by showing the scope and sequence of board-relevant topics through dashboards and the connected assessments. Methods To align USMLE® content to our curriculum, we developed and operationalized a granular, machine‑readable coding library derived from the 2025 USMLE® Content Outline (STEP) and to demonstrate its utility for session‑level mapping and assessment tagging. Results Using a top‑down, left‑to‑right parse of the STEP, we constructed a hierarchical code library with the parent–child relationships preserved. “Normal processes” were assigned 1.0 codes within each system to distinguish physiology/anatomy from pathology. We then used a structured prompt with a generative AI assistant to propose STEP codes for preclerkship session-level learning objectives. The library spans 18 STEP domains and supports alignment at multiple levels of granularity. Early use cases include (1) session‑level mapping to surface gaps and redundancies and (2) assessment tagging to enable exam blueprinting by STEP domain. Conclusions To mitigate the “parallel curriculum,” these linked STEP codes can, in the future, be visualized at a course- or phase-level to provide faculty and students linkages between session-level course objectives and assessments to STEP. This coding library is generalizable across courses, phases, and external institutions and can be mapped by AI technologies.