Developing and Validating a Machine Learning Model to Predict Acute Kidney Injury After Non-Cardiac Surgery with Added Frailty Assessment

What this Trial Is About

Researchers are developing and validating a machine learning model to predict acute kidney injury AKI after non-cardiac surgery in adult patients. The study also aims to assess whether including frailty status as a new factor improves the models ability to predict AKI. This observational study uses data collected from patients undergoing various types of non-cardiac surgeries. The study uses two sets of data retrospective data from adult patients who had non-cardiac surgery between July 2015 and June 2025, and prospective data collected from July 2025 to February 2026. The retrospective data helped develop and optimize the model using multiple machine learning techniques and a wide range of patient and surgical information. The prospective data includes frailty assessments and will be used to validate and update the model. Participants information such as demographics, lab tests, comorbidities, and surgical details are collected from electronic medical records. For the prospective group, frailty status and outcomes like postoperative complications, mortality, hospital stay length, and costs are recorded. The primary outcome measured is acute kidney injury within 7 days after surgery. The study tracks these outcomes during the perioperative period to evaluate the models performance and value of frailty assessment.

A Machine Learning Prediction Model for Postoperative Acute Kidney Injury in Non-Cardiac Surgery Patients