Artificial Intelligence (AI) is increasingly utilized in healthcare as wearable technology, virtual assistants, or to aid decision-making. This study evaluates the feasibility, effectiveness, and limitations of AI-based ChatGPT-4.0 in developing 8-week exercise programs for cases with lumbar disc herniation (LDH), chronic migraine (CM), and urge urological incontinence (UUI). ChatGPT-4.0 was questioned about exercise frequency, intensity, type, duration, targeted muscles, repetitions, theraband strengths, perceived difficulty, and aerobic exercise recommendations. The answers given were evaluated by experts. Expert evaluations determined that ChatGPT-4.0 successfully created literature-based programs for LDH, CM, and UUI, including cervical, lumbar stabilization, and pelvic floor exercises. However, issues arose: theraband resistances and plank-like challenging exercises for LDH were introduced too early, potentially causing rapid progression. In CM, isometric exercises risk triggering attacks, and progression rates were accelerated in all cases. These findings highlight ChatGPT-4.0’s inability to fully adapt programs to patient medical conditions, emphasizing the critical role of physical therapists in designing individualized exercise programs.
This study is a qualitative study and no attempt was made to evaluate or treat the patients. It was only suggested that real patients be included in an exercise program after a physician's diagnosis and how artificial intelligence could prepare these exercise programs according to real patient diagnoses was investigated. Therefore, this study does not need an ethics committee application.
There is no person or organization that financially supports this study.
Primary Language | English |
---|---|
Subjects | Sports Science and Exercise (Other) |
Journal Section | Original Research Articles |
Authors | |
Early Pub Date | February 25, 2025 |
Publication Date | February 28, 2025 |
Submission Date | January 11, 2025 |
Acceptance Date | February 19, 2025 |
Published in Issue | Year 2025 Volume: 11 Issue: 1 |