Reasoning under uncertainty in graduate health education

A scaffolded framework using large language models

Datos Bibliográficos

ID22166282
AutoresCarlos Ugrinowitsch (0000-0001-8547-419X, Hungarian University of Sports Science), Leonardo Lamas (Universidade de Brasília), Cleiton Augusto Libardi (0000-0002-9003-7610, Universidade Federal de São Carlos)
Año2026
Volumen11
Fecha de publicación2026-06-30
Peer ReviewedSí
Open AccessSí
TipoARTICLE
RevistaFrontiers in Education (JOURNAL)
Identificadores de la revistaISSN: 2504-284X • E-ISSN: 2504-284X
EditorialFrontiers Media SA (PUBLISHER • CH)
DOI10.3389/feduc.2026.1857036
OpenAlexW7166725011
IdiomaEN
Referencias citadas32

The management of non-communicable diseases increasingly demands reasoning that integrates multiple mechanisms, disciplinary perspectives, and contextual constraints. Although graduate training in the health sciences is highly effective in helping learners apply stabilized evidence through Evidence-Based Practice (EBP), clinical guidelines, and decision pathways, it often gives less explicit attention to how learners should reason when evidence is incomplete, conflicting, or poorly aligned with the problem at hand. These tools compress scientific knowledge into clinical schemas that, through repeated use, support heuristic decision-making in routine contexts. The challenge emerges when evidence is sparse, conflicting, or misaligned with the clinical problem. At that point, schema-based strategies become insufficient, and uncertainty becomes a structural feature of decision-making rather than a temporary lack of data. This Hypothesis and Theory article proposes a pedagogical framework for training cross-silo integrative reasoning under uncertainty in graduate health education. The framework distinguishes three layers of clinical reasoning and operationalizes the third (i.e., the Uncertainty Layer) through five adaptive simulators that can train both horizontal and vertical integration, in which large language models function as constrained epistemic adversaries. A non-normative rubric, intended for developmental monitoring rather than summative judgment of correctness, assesses how learners deal with uncertainty across simulator sessions. The framework does not replace EBP, clinical expertise, or mentorship; it formalizes the reasoning processes required when stabilized evidence reaches its limits, making expert judgment under uncertainty explicit, inspectable, and teachable within graduate health-sciences education

Clinical judgment · Conceptual framework · Discipline · Graduate students · Heuristic · Summative assessment · Clinical Reasoning and Diagnostic Skills · Innovations in Medical Education · Simulation-Based Education in Healthcare