Artificial Intelligence for Predicting Clinical Outcomes in Interventional Pain Medicine for Spine Disorders: A Systematic Review.

OBJECTIVES

Artificial intelligence (AI) applications are being increasingly explored in pain medicine due to AI's ability to handle multidimensional data and analyze complex, nonlinear relationships. There is a need to identify and understand the advances that have been made in developing AI-based clinical prediction models in interventional pain medicine for spine disorders. Therefore, the purpose of this study is to conduct a systematic review of AI-based prediction models for clinical outcomes in interventional spine medicine with a focus on model field of application, performance, and generalizability.

METHODS

A systematic review evaluating AI-based clinical prediction models in interventional pain medicine for spine disorders was conducted using the PubMed/MEDLINE and Scopus databases in February of 2025. Articles meeting eligibility criteria had standardized data extracted and were assessed for their application, performance (primarily based on area under the receiver operating characteristic curve [AUROC] and accuracy), and generalizability (internal and/or external validation).

RESULTS

A final total of nine studies were included in this systematic review. Of these nine, four of the studies were pertaining to epidural steroid injections and five of the studies were pertaining to spinal cord stimulators. Two studies (22.2%) out of nine achieved an excellent (> 0.90) AUROC or accuracy for their AI-based prediction models. One study (11.1%) externally validated their AI-based prediction model.

DISCUSSION: AI-based clinical prediction models are limited to epidural steroid injections and spinal cord stimulators. Additionally, there is a need to improve model performance and generalizability through external validation prior to clinical translation.