•  
  •  
 

Corresponding Author

Alaa Eldin Mahmoud Mohamed Elsaed

Document Type

Meta Analysis

Abstract

Background: Artificial intelligence (AI) and machine learning (ML) have significant potential for improving clinical decision-making by analyzing complex healthcare data. However, their use in predicting outcomes after shoulder arthroplasty still needs to be explored.

Aim: This study aims to evaluate the impact of AI in predicting the success rates of total and reverse total shoulder arthroplasty.

Methods: We conducted a comprehensive literature search in PubMed, Web of Science, and SCOPUS, focusing on randomized controlled trials and observational studies. Our analysis included various outcomes such as Area Under the Precision-Recall Curve (AUPRC) scores and Mean Absolute Errors (MAE) in predicting various shoulder function scores.

Results: Our meta-analysis included data from 154,988 patients with an average age of 69.63 years. We found the average AUPRC score to be 0.839, indicating robust model performance. The MAEs for various shoulder function scores were as follows: Global Shoulder Function score showed an MAE of 1.025, indicating a high level of prediction accuracy. The VAS pain score prediction had an MAE of 1.00, demonstrating the model's efficacy in pain assessment. The ASES score prediction yielded an MAE of 11.61, while active forward elevation had an MAE of 17.663. Active external rotation was associated with an MAE of 12.771, and the constant score prediction showed an MAE of 9.095.

Conclusion: AI has the potential to revolutionize the field of shoulder arthroplasty, enhancing surgical decision-making and patient outcomes. Despite challenges, AI offers promising avenues for improved orthopedic care.

Keywords

Artificial intelligence; Machine Learning; Total Shoulder Arthroplasty; Reverse Total Shoulder Arthroplasty

Subject Area

Orthopedics

Download

Share

COinS