What is the clinical evidence from this study?

Study design: Observational. Population: Acute coronary syndrome in patients with kidney failure receiving dialysis (n=60491). Intervention: Percutaneous Coronary Intervention (PCI) vs. Coronary Artery Bypass Graft (CABG) and no treatment. Primary outcome: Mortality prediction from 30 days to 5 years.

What question did this study set out to answer?

This research aims to develop a machine learning model to predict mortality outcomes based on revascularization strategies in dialysis patients after acute coronary syndrome.

March 28, 2026Kidney International Reports0 citationsOpen Access

Wcn26-1880 Machine Learning to Predict Outcomes by Revascularization Strategy in Patients With Kidney Failure After Acute Coronary Syndrome

SSSilvi ShahUniversity of Cincinnati HSHanniel ShihUniversity of Cincinnati HCHIMAJA Chintalapalli

Key Result

Percutaneous coronary intervention decreased predicted mortality more than CABG at 30 days (6.48% vs 2.33%) and 5 years (3.45% vs 3.30%) in an AI-based model of dialysis patients with ACS.

Key Points

This research aims to develop a machine learning model to predict mortality outcomes based on revascularization strategies in dialysis patients after acute coronary syndrome.
Utilized US Renal Data System to gather clinical and demographic data from 60,491 dialysis patients.
Performed stratified random sampling with an 80-20 split and three-fold cross-validation.
Trained and evaluated five machine learning models and one deep learning model for mortality prediction.
Measured model performance using precision, recall, F1 score, and AUROC.
The gradient-boosting model showed AUROC between 0.6532 and 0.7898, outperforming other models.
The deep neural network achieved the highest F1 score of 0.9299 for 5-year prediction.
PCI reduced predicted mortality by 6.48% at 30 days and 3.45% at 5 years; CABG reduced mortality by 2.33% and 3.30%, respectively.
In both genders, PCI produced the most significant decrease in predicted mortality over the years.

Study Design

Type

Observational (n=60,491)

Multicenter

Yes

Structured PICO

Does PCI reduce mortality compared to CABG in dialysis patients with acute coronary syndrome?

Population

Dialysis patients with acute coronary syndrome

Intervention

Percutaneous Coronary Intervention (PCI)

Comparator

Coronary Artery Bypass Graft (CABG) or no treatment

Outcome

Mortality from 30 days to 5 yearshard clinical

A deep neural network model predicts that PCI has a greater effect on reducing mortality compared to CABG from 30 days to 5 years in dialysis patients with acute coronary syndrome.

Abstract

Introduction: Due to retrospective data and the exclusion of patients with kidney failure in clinical trials, the ability to predict mortality in patients with acute coronary syndrome (ACS) receiving dialysis is limited, particularly in females, by the type of revascularization (Percutaneous Coronary Intervention PCI and Coronary Artery Bypass Graft CABG).Methods: Using the United States Renal Data System, 51 clinical and demographic features were extracted from 60,491 dialysis patients with acute coronary syndrome.The performance of an artificial intelligencebased mortality prediction model in dialysis patients by revascularization strategy type and sex from 30 days to 5 years after the first acute coronary syndrome hospitalization, based on precision, recall, F1 score, and area under the receiver operating curve (AUROC).Stratified random sampling was employed with an 80-20 split and a 3-fold crossvalidation.Five machine learning models and one deep learning model were trained and evaluated.Results: The gradient-boosting machine learning model outperformed other models with an AUROC between 0.6532 and 0.7898.The deep neural network had the highest F1 score, achieving 0.9299 at the 5-year prediction.At 30 days, PCI decreased predicted mortality by 6.48%, while CABG decreased predicted mortality by 2.33%.At 5 years, PCI decreased the predicted mortality by 3.45%, while CABG decreased the predicted mortality by 3.30%.In both males and females, from 30 days to 5 years, PCI led to the greatest decrease in predicted mortality, while no treatment led to the greatest increase in predicted mortality. Conclusion:This study is the first to develop an artificial intelligencebased mortality prediction model for patients with acute coronary syndrome undergoing dialysis.The deep neural network showed the best overall performance.PCI had a greater effect on reducing predicted mortality as compared to CABG from 30 days to 5 years in both males and females with acute coronary syndrome undergoing dialysis.I have potential conflict of interest to disclose.

Ask AI

Helpful

Bookmark

View Full Paper