Clinical Scorecard: Utilization of Natural Language Processing and Large Language Models in Evaluating Sports Injuries and Making Rehabilitation Decisions: A Scoping Review

At a Glance

Key Highlights

• 27 studies included, primarily from the United States (37.0%)
• Main research types were algorithm evaluation and benchmarking
• Challenges include readability issues and AI hallucinations
• Overall compliance rate with MINIMAR report quality assessment was 61.70%
• LLMs show potential for personalized clinical recommendations and exercise prescriptions

Guideline-Based Recommendations

Diagnosis

• Utilize NLP and LLMs to analyze unstructured text data in sports injury assessment

Management

• Incorporate AI technologies for generating personalized rehabilitation decisions

Monitoring & Follow-up

• Assess the methodological characteristics and clinical efficacy of NLP applications

Risks

• Address challenges related to readability and AI hallucinations in clinical applications

Patient & Prescribing Data

Professional athletes and recreational exercisers with sports-related injuries

Focus on improving model reliability and bridging geographic gaps in research

Clinical Best Practices

• Follow established standards such as MINIMAR for reporting AI algorithms
• Enhance the transition from baseline assessment to clinical application

Related Resources & Content

• Joanna Briggs Institute
• PRISMA-ScR guidelines
• Open Science Framework