Keywords
Association for Athletic Training Education 2025 Symposium
Abstract
Introduction: Rigorous accreditation standards require students to demonstrate competency in various clinical proficiencies through clinical practice or simulation-based educational activities. Therefore, simulation has become an integral component of athletic training education, as scenarios can provide realistic and reproducible experiences for all students. To that end, our study had two primary objectives: investigate the ability of artificial intelligence (AI) to generate accurate and realistic athletic training simulation scenarios and evaluate AI's effectiveness in teaching and reinforcing the emergency assessment process through student-created scenarios. Methods: Phase one, AI developed three different scenarios for patients experiencing (1) an ovarian cyst, (2) a hypoglycemic event, and (3) deep vein thrombosis (DVT). Three content expert reviewers assessed the scenarios using the Simulation Validation Checklist.2 Scenarios had to meet the Healthcare Simulation Standards of Practice, the Commission on Accreditation of Athletic Training Education Standards, and be appropriate for master's level athletic training students. Experts (N=10) evaluated the scenarios for realism, accuracy, and completeness using a 25-question Likert-scale and open-ended responses. Phase two included students enrolled in the Emergency Care of Athletic Injuries course (N=14). Students used an AI platform to generate emergency care scenarios for athletic populations aged 16 to 25. Scenarios had to be realistic and of intermediate complexity. Lastly, students participated in low-fidelity simulations and reflected on their performance. Descriptive statistics were performed using JASP (Version 0.17.2). Frequencies and percentages were used to describe the Likert scale data, which scored the simulations and attributes on a 5-point scale (1=extremely unrealistic, 2=unrealistic, 3=realistic, 4=very realistic, 5=extremely realistic). Open response data were grouped by themes and described narratively. Results: Findings from the first phase indicated that while ChatGPT was effective at creating the foundation for a medical simulation, scenarios were often disorganized and incomplete. Scenarios lacked complete, accurate patient history and background information. Details such as vital signs, expected signs and symptoms, and patient disposition were missing or inaccurate, affecting believability. The highest scoring scenario attribute was patient profile (93/150), followed by history of present illness (92/150), past medical history (90/150), and realism (81/150). Among the three scenarios, hypoglycemia performed best (134/150), followed by ovarian cyst (128/150), with DVT scoring lowest (94/150). The second phase evaluated students' knowledge and skills working through the emergency evaluation process including rapid trauma assessment and managing critical conditions. Student feedback when compared to the AI assignment, underscored the necessity for additional practice in information gathering, particularly in evaluating subtle variations in vital signs and breath sounds. Students who reported needing more practice assessing vitals and interpreting findings had lower scores for realism, completeness, and accuracy on their AI assignment. Conversely, students who reported feeling more confident had higher scores on the scenario creation grading rubric. Translation to Practice: AI in athletic training education provides an opportunity to enhance the learning process by allowing students to actively engage in the simulation development process. However, it is important to recognize AI's limitations such as its inability to replicate the human component of the decision-making process. Creating simulations reinforces didactic content through the creation of life-like conditions commonly seen in sports. Assisted by AI, students selected a condition such as sudden cardiac arrest or open fracture. They methodically work through developing their scenario guided by content from class. Ensuring a high level of realism forces students to think through all plausible conditions and treatments to effectively manage their chosen condition. Performance inconsistencies between the AI assignment and end of semester simulation identify gaps in classroom instruction, students understanding, and areas for course improvement.
Recommended Citation
Cobble, D; Schroeder, L; Vaughn, J; and Ford, S
(2025)
"Using AI-Generated Scenarios to Enhance Learning of The Emergency Assessment Process,"
Clinical Practice in Athletic Training: Vol. 8:
Iss.
1, Article 5.
Available at:
https://scholars.indianastate.edu/clinat/vol8/iss1/5
Included in
Biology Commons, Body Regions Commons, Diagnosis Commons, Exercise Physiology Commons, Exercise Science Commons, Health and Physical Education Commons, Investigative Techniques Commons, Kinesiotherapy Commons, Medical Physiology Commons, Movement and Mind-Body Therapies Commons, Musculoskeletal, Neural, and Ocular Physiology Commons, Musculoskeletal System Commons, Occupational Therapy Commons, Other Kinesiology Commons, Other Physiology Commons, Other Rehabilitation and Therapy Commons, Physical Therapy Commons, Physiological Processes Commons, Physiotherapy Commons, Recreational Therapy Commons, Research Methods in Life Sciences Commons, Sports Sciences Commons, Therapeutics Commons, Tissues Commons
