A Systematic Review of Artificial Intelligence in Orthopaedic Disease Detection: A Taxonomy for Analysis and Trustworthiness Evaluation

Abstract Orthopaedic diseases, which affect millions of people globally, present significant diagnostic challenges, often leading to long-term disability and chronic pain. There is an ongoing debate across the literature regarding the trustworthiness of artificial intelligence (AI) in detecting orthopaedic diseases. This systematic review aims to provide a comprehensive taxonomy of AI applications in orthopaedic disease detection. A thorough literature search was conducted across five major databases (Science Direct, Scopus, IEEE Xplore, PubMed, and Web of Science) covering publications from January 2019 to 2024. Following rigorous screening on the basis of predefined inclusion criteria, 85 relevant studies were identified and critically evaluated. For the first time, this review classifies AI contributions into six key categories of orthopaedic conditions on the basis of medical perspective: arthritis, tumours, deformities, fractures, osteoporosis, and general bone abnormalities. In addition to analyzing motivations, challenges, and recommendations for future research, this review highlights the various AI techniques employed, including deep learning (DL), machine learning (ML), explainable AI (XAI), fuzzy logic, and multicriteria decision-making (MCDM), as well as the datasets utilized. Furthermore, the trustworthiness of AI models is evaluated on the basis of seven AI trustworthiness components, aligned with European Union guidelines, within each category. These findings underscore the need for high-quality research to ensure that AI computational systems in orthopaedic disease detection are reliable, safe, and ethical. Future research should focus on optimizing AI algorithms, improving dataset diversity, and addressing ethical and regulatory challenges to ensure successful integration into clinical practice.