The item generation methodology of an empiric simulation project

Area of Science:

• Medical Education Technology
• Computer-Based Testing
• Clinical Simulation

Background:

• The American Board of Family Practice (ABFP) is developing a novel computer-based testing system.
• The system aims to create realistic clinical encounters for assessment purposes.

Purpose of the Study:

• To describe the development of a computer-based testing system for family practice.
• To outline the methodology for generating realistic and variable simulated patient encounters.
• To discuss the potential for cost-effective content maintenance and scalability.

Main Methods:

• Utilizes an item generation process to create simulated patients with stochastically produced life histories.
• Employs constraint mechanisms, including Bayesian networks and fuzzy definitions, to ensure plausible simulations and introduce variability.
• Examinees interact via queries and interventions, with scoring based on multiple management plans.

Main Results:

• The system generates realistic and variable patient simulations from a knowledge base.
• Bayesian networks effectively manage dependencies between medical concepts.
• Reusable components (Health States, "Lead To" objects, Plans) enhance system flexibility.

Conclusions:

• The computer-based testing system offers a robust framework for assessing clinical skills.
• Despite high initial knowledge acquisition costs, the system promises low maintenance costs and economies of scale through component reuse.
• This approach has the potential to significantly advance medical education and assessment methodologies.