Surgical site infection (SSI) remains the most common complication after pediatric surgery, prolonging recovery and increasing cost1, 2, 3, 4. Two fundamental challenges persist: predicting who will develop an SSI and detecting infections promptly once they occur. Traditional prediction methods rely on broad wound classification and limited patient-specific variables5. Detection systems depend on manual chart reviews, requiring clinicians to search across laboratory results, imaging, and free-text documentation.

The advent of electronic health records (EHRs) has expanded the scope of available patient data, combining discrete and unstructured elements6, 7, 8. Even basic automated EHR workflows, such as flagging antibiotic use, microbial cultures, and infectious disease consultations, can reduce the surveillance burden and improve case detection accuracy9.

Artificial Intelligence has been rapidly incorporated into pediatric surgical care10, and has begun to address both of these challenges. Machine learning (ML), Natural Language Processing (NLP), and Large Language Models (LLMs) now offer the ability to integrate structured and unstructured data into powerful predictive and surveillance models11, 12, 13.

This review synthesizes current evidence, identifies limitations, and outlines future directions for AI in pediatric SSI prediction and detection.