Predicting Postoperative Outcomes in Lower Gastrointestinal Surgery: A Machine Learning Approach Using Electronic Health Records

As surgical care becomes more complex, artificial intelligence offers promising opportunities to improve patient outcomes. This study investigated the use of machine learning to predict postoperative complications and the length of hospital stay in patients undergoing lower gastrointestinal surgery. We analyzed the operative electronic health records of 771 patients using Support Vector Machine models. The results indicate that patients' preoperative factors, such as age and the number of diagnoses, are significant predictors of postoperative complications. Precisely, an increased number of diagnoses correlates with increased complications and an elongated length of hospital stay. The Support Vector Machine model achieved a cross-validation accuracy of 75.56% in complication prediction. Integrating predictive models into practice can support early risk identification and targeted care. These findings highlight the potential of artificial intelligence in perioperative care and underscore the need for additional research to refine predictive models in nursing practice.