Presented at the Military Health System Research Symposium (MHSRS), Kissimmee FL (August 2017)
Background
One commonly used approach to training trauma assessment and other emergency medicine skills is simulation-based training, allowing trainees to practice their skills with simulated patient scenarios. This approach often uses standardized metrics and methods to evaluate those metrics to assess performance and provide feedback. Previous research has defined and validated objective measures of performance using a high-fidelity manikin-based simulation (Holcomb et al., 2002). This paper describes the design decisions and development of a lower-fidelity, more easily deployed and reused virtual patient with automatic, objective metrics to practice and assess trauma assessment skills. A previously reported hierarchical analysis of trauma assessment skills informed the development of the task environment and the automatically assessed metrics (Bauchwitz et al., 2016).
Methods
We performed an analysis of existing instructional materials, such as EMT-B, ATLS and TNCC, and conducted cognitive task analysis of the cognitive skills required for primary and secondary trauma assessment. Based on these analysis we identified 21 common and critical injuries for training, and designed twelve scenarios to provide clinically realistic combinations of those injuries. We identified four key metrics for assessing performance: (1) the correctness of the actions to find and treat injuries; (2) the correctness of the order of treatment; (3) the time to treat; and (4) the harm done to patients. Based on the injuries, scenarios, and assessment methods, we designed and developed a tablet-based interface modeled on a previously developed simulation for combat lifesaver skills (Hobbs et al., 2012). Working with subject matter experts we developed a state-based patient simulation with lower-fidelity physiology to enable trainees to diagnose and treat the virtual patient. Transitions in the patient state can be observed by trainees and recorded to calculate metrics. Automatic scoring mechanisms were developed to track the correctness of the actions to find and treat injuries, the correctness of the order of treatment, and the time to treat. Scoring algorithms were designed to be flexible, so different training sites or different curricula are able to adjust scoring metrics by changing only parameters. For example, scoring parameters can be changed to emphasize treating massive hemorrhage first, or securing the airway first, depending on the organization and curriculum.
Results
We have successfully identified and implemented three objective and automatic metrics for trauma assessment in virtual simulation. We have implemented five scenarios to demonstrate the capabilities of the system, along with automatic scoring mechanisms that log and report a cumulative score sheet per performance. Scenario configuration and scoring are data-driven (from human-readable files), meaning authors can change scenario data, patient state, and scoring metrics without reprogramming the simulation. We have demonstrated the system to end users, and informally assessed usability.
Conclusions
Virtual simulation can effectively assess and enable practice trauma assessment, reducing the costs of training and increasing the accessibility of training these critical skills. Future work remains to assess training effectiveness and compare virtual and manikin-based simulations.
1 Charles River Analytics
2 UMass Memorial Medical Center/UMass Medical School
3 Nurse Corps, US Army Reserves
4 The Pennsylvania State University
For More Information
To learn more or request a copy of a paper (if available), contact Peter Weyhrauch.
(Please include your name, address, organization, and the paper reference. Requests without this information will not be honored.)
