Machine Learning Models for Prediction of Acute Kidney Injury After Noncardiac Surgery

Rao Sun

Status: Completed

Conditions: Postoperative Acute Kidney Injury

Treatments: Other: no intervention

Study type: Observational

Funder types: Other

Identifiers

NCT06146829

TJH-20230608C

Details and patient eligibility

About

Acute kidney injury (AKI) is a common surgical complication characterized by a rapid decline in renal function. Patients with AKI are at an increased risk of developing chronic kidney disease and end-stage renal disease, which has been associated with an increased risk of morbidity, mortality and financial burdens. Identifying high-risk patients for postoperative AKI early can facilitate the development of preventive and therapeutic management strategies, and prediction models can be helpful in this regard.

The goal of this retrospective study is to develop prediction models for postoperative AKI in noncardiac surgery using machine learning algorithms, and to simplify the models by including only preoperative variables or only important predictors.

Enrollment: 88,367 patients

Sex: All

Ages: 18+ years old

Volunteers: No Healthy Volunteers

Inclusion criteria

• Adult patients (age ≥ 18 years) who had a serum creatinine measurement within 10 days before surgery and at least one measurement within 7 days after surgery.
• Eligible surgeries encompassed general, thoracic, orthopedic, obstetric, gynecology, and neurosurgery procedures lasting longer than 1 hour

Exclusion criteria

• Patients with concurrent cardiac, vascular, urological, or transplant surgeries.
• Patients with an American Society of Anesthesiologists (ASA) physical status V.
• Patients with end-stage renal disease (i.e., a glomerular filtration rate [eGFR] of 15 mL/min/1.73 m² or receiving hemodialysis).