Prediction Model for Postoperative Acute Kidney Injury in Patients Undergoing Lung Transplantation Using Machine Learning: a Retrospective Cohort Study

Since the first report of lung transplantation in humans in 1963, there has been rapid progress in both the quantity and quality of lung transplantation, driven by the significant advancements in immunosuppressive agents since the mid-1990s. In 2004, a total of 1,815 lung transplantations were reported worldwide. Patients undergoing lung transplantation are often elderly and face not only impaired lung function but also overall health instability, leading to the potential occurrence of respiratory complications post-surgery, even with successful lung transplantation outcomes.

Postoperative acute renal injury (AKI) can result in temporary or even chronic renal dysfunction. AKI following surgery can lead to an increase in hospitalization duration, complications, and the need for additional treatment. Various factors are associated with postoperative renal dysfunction after lung transplantation, including sustained hypoperfusion, hypoperfusion related to intraoperative and postoperative bleeding, heart failure, acute myocardial infarction, pulmonary embolism, sepsis, and more. Medications related to renal dysfunction include those associated with thrombosis or embolism, such as aminoglycosides, amphotericin B, non-steroidal anti-inflammatory drugs (NSAIDs), proton-pump inhibitors, contrast agents, and others. Additionally, graft-versus-host disease is known to be related to renal dysfunction.

The retrospective analysis of medical records from adult patients who underwent lung transplantation aims to investigate patient characteristics, anesthesia methods, intraoperative tests, and the occurrence of postoperative acute renal dysfunction. The goal is to analyze the incidence and risk factors of postoperative renal dysfunction and develop a predictive model through machine learning.