Utility of Machine Learning, Natural Language Processing, and Artificial Intelligence in Predicting Hospital...

Area of Science:

• Orthopaedic Surgery
• Medical Informatics
• Data Science

Background:

• Hospital readmissions after orthopaedic surgery are a significant concern, prompting research into predictive factors.
• Existing strategies aim to identify risk factors and causative agents for readmissions.
• The role of advanced computational methods in predicting these readmissions requires systematic evaluation.

Purpose of the Study:

• To systematically review and summarize the literature on the use of artificial intelligence (AI), natural language processing (NLP), and machine learning (ML) for predicting hospital readmissions after orthopaedic and spine surgeries.
• To assess the predictive performance of these computational tools.

Main Methods:

• A systematic review and meta-analysis was conducted, searching PubMed, Embase, and Google Scholar up to August 30, 2023.
• Studies utilizing AI, NLP, and ML for predicting readmission rates after orthopaedic procedures were included.
• Data on study design, patient population, models used, predictors, and accuracy (C-statistic) were extracted and analyzed.

Main Results:

• Twenty-six studies were included, with a mean C-statistic of 0.71, indicating reasonable predictive capability.
• Spine surgeries were most frequently studied (57%), but hip/knee arthroplasty models showed higher accuracy (mean C-statistic = 0.79).
• Models incorporating intraoperative/postoperative data and single-institution data performed better; however, most studies had a high risk of bias.

Conclusions:

• AI, NLP, and ML tools demonstrate reasonable performance in predicting readmissions post-orthopaedic surgery.
• Standardization of study methodologies and improved data analysis are crucial for enhancing the reliability of predictive models.
• Future research should focus on these areas to yield more robust and actionable findings.