โก Quick Summary
This study evaluated the feasibility and safety of AI-generated exercise prescriptions for at-risk populations, utilizing a large language model (Gemini 2.5). The findings suggest that while these AI-generated prescriptions show potential, expert involvement remains crucial for effective implementation.
๐ Key Details
- ๐ง Technology Used: Large language model (Gemini 2.5, Google LLC)
- ๐ฉโโ๏ธ Expert Evaluation: Conducted by a panel of experts using a structured rubric
- ๐ Assessment Criteria: Safety, feasibility, guideline alignment, and personalization
- ๐ Inter-expert Agreement: Low (ICC (2,3) = 0.139)
- ๐ Internal Consistency: High (Cronbach’s alpha > 0.92)
๐ Key Takeaways
- ๐ค AI-generated exercise prescriptions can serve as supportive decision-making tools.
- ๐ฉโโ๏ธ Expert involvement is essential for ensuring safety and effectiveness.
- ๐ Low inter-expert agreement indicates variability in expert evaluations.
- ๐ Improved scores in safety and guideline alignment were noted with better prompt structuring.
- ๐ Further structuring did not consistently enhance outcomes.
- ๐งฉ Personalization remains a critical factor in exercise prescription.
- ๐ Study published in the Journal of Clinical Medicine.
- ๐ PMID: 41899378.
๐ Background
The integration of artificial intelligence in healthcare is rapidly evolving, particularly in exercise science and sports medicine. The potential for AI to generate personalized exercise programs could revolutionize how we approach fitness and rehabilitation, especially for at-risk populations. However, the practical applicability of these AI-generated prescriptions has yet to be thoroughly validated in complex clinical contexts.
๐๏ธ Study
This study aimed to assess the practical utility of AI-generated exercise prescriptions under expert supervision. The researchers analyzed outputs from the Gemini 2.5 model, applying three levels of prompt structuring to evaluate the generated exercise prescriptions. Experts then assessed these outputs based on a structured rubric that focused on safety, feasibility, guideline alignment, and personalization.
๐ Results
The results indicated that while AI-generated exercise prescriptions exhibited a certain level of structural completeness, the inter-expert agreement was notably low, with an ICC of 0.139. In contrast, the expert-specific internal consistency was high, with a Cronbach’s alpha exceeding 0.92. Notably, improvements in safety and guideline alignment were observed with enhanced prompt structuring, although additional structuring did not consistently yield further benefits.
๐ Impact and Implications
The findings of this study highlight the potential of AI-generated exercise prescriptions as valuable tools in clinical decision-making. However, the low inter-expert agreement underscores the need for expert involvement in the process. As AI continues to evolve, integrating these technologies into clinical practice could enhance personalized care for at-risk populations, provided that expert oversight is maintained.
๐ฎ Conclusion
This study demonstrates the promising potential of AI in generating exercise prescriptions, but it also emphasizes the importance of expert evaluation in ensuring safety and effectiveness. As we move forward, further research is needed to refine these AI tools and explore their practical applications in diverse clinical settings. The future of exercise prescription could be significantly enhanced by the collaboration between AI technologies and healthcare professionals.
๐ฌ Your comments
What are your thoughts on the use of AI in exercise prescriptions? Do you believe that AI can effectively support healthcare professionals in this area? ๐ฌ Share your insights in the comments below or connect with us on social media:
AI-Generated Exercise Prescriptions for At-Risk Populations: Safety and Feasibility of a Large Language Model Assessed by Expert Evaluation.
Abstract
Background/Objectives: In exercise science and sports medicine, the potential use of large language models for generating personalized exercise programs is being explored. However, the practical applicability of AI-generated exercise prescriptions has not yet been sufficiently validated, particularly in complex clinical contexts. This study aimed to evaluate their practical utility under expert supervision. Methods: Exercise prescription outputs generated by a large language model (Gemini 2.5, Google LLC) were analyzed using clinical cases incorporating complex exercise-related considerations. Three levels of prompt structuring were applied. Experts evaluated the outputs using a structured rubric assessing safety, feasibility, guideline alignment, and personalization. Inter-expert agreement was assessed using intraclass correlation coefficients (ICC), and expert-specific internal consistency was evaluated using Cronbach’s alpha. Results: AI-generated exercise prescriptions demonstrated a certain level of structural completeness. However, inter-expert agreement was low (ICC (2,3) = 0.139), whereas expert-specific internal consistency was high (Cronbach’s alpha > 0.92). Prompt structuring from Stage 1 to Stage 2 was associated with improved mean scores in safety and guideline alignment. Additional structuring did not consistently yield further improvements. Conclusions: AI-generated exercise prescriptions may have practical potential as supportive decision-making tools when expert involvement is assumed. Nonetheless, expert judgments did not converge toward a single evaluative standard, reflecting the inherently expert-dependent nature of exercise prescription.
Author: [‘Choi M’, ‘Park J’, ‘Lee M’, ‘Beom J’, ‘Jung SY’, ‘Lee K’]
Journal: J Clin Med
Citation: Choi M, et al. AI-Generated Exercise Prescriptions for At-Risk Populations: Safety and Feasibility of a Large Language Model Assessed by Expert Evaluation. AI-Generated Exercise Prescriptions for At-Risk Populations: Safety and Feasibility of a Large Language Model Assessed by Expert Evaluation. 2026; 15:(unknown pages). doi: 10.3390/jcm15062457