•  
  •  
 

Stony Brook Orthopaedic and Musculoskeletal Journal

Keywords

Artificial Intelligence, Patellofemoral Instability, Pediatric, Knee, MPFL

Abstract

Introduction: The purpose of this study was to elucidate if artificial intelligence (AI) can generate novel research hypotheses regarding patellofemoral instability (PFI) in the pediatric population that have not previously been investigated. We hypothesized that AI would be able to generate new hypotheses that have not yet been presented or published.

Methods: In December of 2023, Chat Generative Pre-trained Transformer-3.5 (ChatGPT) was used to create a list of 25 research hypotheses about PFI in both the adult and pediatric population. The prompt used to generate the list of hypotheses was: “List 25 detailed and reproducible research ideas regarding Patellofemoral Instability for both the adult and pediatric population”. The advanced search option under PubMed was then used to search for articles pertaining to Pediatric PFI between January 2022 and December 2023. The terms: Patellar Instability, Pediatric and Pediatric Patella Instability were used to conduct the search. Additionally, the American Association of Orthopaedic Surgeons (AAOS), American Orthopaedic Society of Sports Medicine (AOSSM) and Pediatric Orthopaedic Society of North America (POSNA) 2022 and 2023 meetings were searched for abstracts and presentations pertaining PFI. Identified articles and abstracts were then reviewed and labeled as matching or not matching each hypothesis generated by Chat GPT. Subsequently, each study was given an originality score from 0-5, with 0 meaning the study matched more than one hypothesis and 5 meaning it matched no hypotheses.

Results: There were 52 articles and abstracts reviewed, of which 49 met the inclusion criteria. The mean originality score was 1.29 ± 1.49. There were 17 studies with a score of 0, 21 with a score of 1, 4 with a score of 2, 2 with score of 3, 2 with score of 4, and 4 with a score of 5. Of the 25 hypotheses generated by ChatGPT, 13 hypotheses had no studies that matched, and 3 hypotheses had 10 or more studies match.

Conclusion: We find that ChatGPT was capable of generating several novel research hypotheses regarding PFI in the pediatric population. Artificial intelligence platforms can serve as a great starting point for brainstorming research ideas.

Disclosure Statement

The authors have nothing to disclose.

Corresponding Author

James Barsi, MD

james.barsi@stonybrookmedicine.edu

Download

Share

COinS