This is an open access article distributed under the terms of the Creative Commons Attribution License, which permits unrestricted use, distribution, reproduction and adaptation in any medium and for any purpose provided that it is properly attributed. For attribution, the original author(s), title, publication source (PeerJ Preprints) and either DOI or URL of the article must be cited.
Introduction: Remediating, or preferably, predicting which residents will have difficulty before they need remediating, is a challenging task. Most of us perform better when pumped for an exam. But how do we respond when under routine pressures? Do weaker learners adapt differently, despite coaching? Methods: Using an adaptation of virtual patient software, we explored how learners cope with handling repetitive yet time-sensitive routine tasks. We emulated the performance of routine tasks within a virtual electronic medical record (EMR) environment, tracking individual learner activity and decision pathways, time to act (with and without enforced pressure from programmed time-outs) and their adaptation trajectories over time with coaching. Learners were assessed using Situational Judgement and modified Script Concordance Testing, with reproducible and granular time constraints introduced into the clinical reasoning process. Results: Our case designs introduce a number of competing elements: time pressures, competing priorities and instructions, resource availability and unpredictable outcomes. Learner behaviour is assessed using a variety of metrics including time-stamped decision points, decision pathways and internal counter scores. Clinical reasoning pathways, as compared to a reference peer panel, are in turn compared with and without the time pressures. Conclusions: Predictive analytics have made great promises in diagnosing problems for learners in difficulty but are complex and expensive to deploy widely. Our simpler, rapidly reproducible approach may provide a more practical solution.
Some final data analysis still pending around ROC data.