Current applications of artificial intelligence in shoulder surgery

Abstract

Introduction: The integration of artificial intelligence (AI) and machine learning (ML) is transforming orthopedic surgery. In shoulder pathology, anatomical variability and the technical complexity of arthroscopic and reconstructive procedures limit standardization, positioning AI as a viable alternative to optimize clinical decision-making.

Objective: To synthesize contemporary scientific evidence regarding the clinical applications, diagnostic accuracy, and predictive capacity of artificial intelligence in shoulder pathology and surgery.

Methods: A review of indexed scientific literature was conducted (period 2022–2025). Eleven original and review articles focusing exclusively on the shoulder joint were selected and analyzed. Variables evaluated included the type of technology utilized (deep learning, ML algorithms, and large language models [LLMs]), diagnostic modality (magnetic resonance imaging [MRI] and arthroscopic video), and the clinical outcomes analyzed.

Results: Deep learning neural networks demonstrated high accuracy in the automated classification of supraspinatus and superior labrum anterior-posterior (SLAP) lesions. Intraoperatively, algorithms objectively quantified arthroscopic field clarity in the presence of bleeding, increasing interobserver diagnostic concordance. Predictive models (such as the CLEARER algorithm) effectively anticipated recurrence following Bankart repair and functional prognosis in rotator cuff repairs. LLMs demonstrated a progressive evolution toward generating reliable medical information.

Conclusions: Artificial intelligence in shoulder pathology fosters a transition toward precision medicine by improving diagnostic imaging, optimizing intraoperative assistance, and refining postoperative prognosis.