Comprehensive Summary
Following robot-assisted radical prostatectomy (RARP) surgery, doctors noticed the development of inguinal hernia (a bulge formation in the groin because of tissue pushing into weak muscle). This complication was identified among patients 1-year post RARP surgery and with no expectations/predictions of the complication foreseen. Therefore in an effort to mitigate the development of inguinal hernias, researchers took a sample of 652 patients that underwent the RARP surgery between 2021 and 2023 and tested out various ML (machine learning) models to see if AI technology could predict the patients that are at high-risk of hernia development. They discovered that XGBoost was the better-performing model that was able to separate out high-risk and low-risk patients the best (AUC ~0.8). Furthermore, with a method called SHAP, doctors were able to recognize the top 5 risk factor predictions, (cancer stage, age, history of abdominal surgery, BMI, and pre-surgery albumin levels), and have now become better equipped to make early predictions about possible developments of hernias in patients within the first year after surgery.
Outcomes and Implications
The findings showcase how machine learning and AI can improve the practice of precision oncology by allowing for more accurate risk classification and individualized postoperative surveillance. If integrated in various multi-center trials, this approach may enable oncologists to better identify high-risk patients early, customize follow-up appointments based on diagnosis intensity, and maybe gain the ability to act sooner before recurrence worsens a patient's condition. AI has the capacity to uncover hidden imaging biomarkers and integrate them with clinical considerations, resulting in a more effective decision-support tool than previous techniques.