AI Model Accurately Predicts Continuous Renal Replacement Therapy Survival

Although CRRT has been in use for many decades, there is still no universally accepted set of clinical guidelines for physicians to determine when to initiate CRRT to ensure a positive outcome. The decision to commence CRRT typically relies on a physician's evaluation of the patient’s medical history, vital signs, lab results, and medications. Given the severe illness of these patients, there is always a degree of uncertainty about their survival during or after the treatment. It is estimated that around 50% of adults who undergo CRRT do not survive, rendering the treatment potentially futile for these patients and their families. Now, a new machine-learning model has been developed that can accurately predict the short-term survival of patients undergoing CRRT.

The machine-learning model developed by researchers at the University of California, Los Angeles (UCLA, Los Angeles, CA, USA) helps doctors decide whether a patient should start CRRT by using data from thousands of patient electronic health records to predict the likelihood of survival following CRRT. Unlike previous models that only predict in-hospital mortality after the initiation of CRRT, this innovative tool provides clinicians with information on whether to start CRRT at all.

The model offers a data-driven tool to aid clinical decision-making. It uses advanced machine-learning techniques to sift through a large and complex array of patient data, a task that has traditionally been challenging for doctors. The study illustrates the potential of integrating machine-learning models into healthcare, enhancing treatment effectiveness, and optimizing the use of medical resources.

“CRRT is often used as a last resort, but many patients do not survive it, leading to wasted resources and false hope for families,” said Dr. Ira Kurtz, chief of the UCLA Division of Nephrology and the study’s senior author. “By making it possible to predict which patients will benefit, the model aims to improve patient outcomes and resource use, by serving as a basis for testing its utility in future clinical trials. Like all machine learning models, it needs to be tested in the real world to determine whether it is equally as accurate in its predictions in patients that it wasn’t trained on.”

Related Links:
UCLA