Improving assessment with virtual patients

AVPs can become valuable components in high stakes medical exams, particularly in later years of courses, however this requires application of established assessment principles to AVP design, distinguishing between linear and branched models, choice and consequence driven designs.

Abstract

Assessments should accurately predict future performance in a wide variety of settings yet be feasible to conduct. In medical education a robust and comprehensive system of assessment is essential to protect the public from inadequate professionals. The parameters for devising such an assessment are well-defined, and good practice for writing examinations well-established. However even excellent written assessments are limited in their predictive validity, and limited in sampling, face and construct validity. The increasing availability and power of computing has led to growing interest in computer simulations for use in examinations, creating assessment virtual patients (AVPs). They can potentially test knowledge and data interpretation, incorporate images, sound or video and test decision making. Such AVPs could represent the most comprehensive, integrated assessment possible that is both objective and feasible. This article focuses on AVP design, distinguishing between linear and branched models, choice and consequence driven designs. It reviews the use of AVPs in the context of assessment theory. It presents different AVP designs discussing their benefits and problems. AVPs can become valuable components in high stakes medical exams, particularly in later years of courses. However this requires application of established assessment principles to AVP design.