Machine Learning Approach to Predict Postoperative Pain and Opioid Usage in Elective Primary Spine Surgery: A...

Area of Science:

• Anesthesiology and Pain Management
• Artificial Intelligence in Medicine
• Spine Surgery Outcomes

Background:

• Investigating the impact of erector spinae plane (ESP) block on pain and opioid use post-spine surgery.
• Utilizing machine learning (ML) to predict outcomes in elective spine surgery patients.

Purpose of the Study:

• To develop and evaluate machine learning models for predicting postsurgical pain and opioid consumption.
• To assess the association of ESP block with pain scores and opioid consumption in spine surgery.

Main Methods:

• Retrospective analysis of 2796 spine surgery cases, divided into control and ESP block groups.
• Development of artificial intelligence (AI) predictive models using gradient boosting ensemble trees.
• Quantification of feature importance and partial dependence analysis for predictor-outcome relationships.

Main Results:

• Univariate analysis indicated a statistically significant but clinically irrelevant 1% increase in pain with ESP block (p=0.01).
• ESP block showed a clinically relevant, though statistically non-significant, 6.7% reduction in opioid consumption (p=0.13).
• The best AI model for predicting average postsurgical pain achieved a mean absolute error of 1.24 on a 10-point scale.

Conclusions:

• Machine learning is feasible for identifying risk factors and predicting population-level pain and opioid consumption in spine surgery.
• The role of ESP block in spine surgery remains uncertain; findings are exploratory associations.
• AI models are intended for population-level insights, not individual patient prediction.