Machine learning models for prediction of adverse events after percutaneous coronary intervention

Abstract

An accurate prediction of major adverse events after percutaneous coronary intervention (PCI) improves clinical decisions and specific interventions. To determine whether machine learning (ML) techniques predict peri-PCI adverse events [acute kidney injury (AKI), bleeding, and in-hospital mortality] with better discrimination or calibration than the National Cardiovascular Data Registry (NCDR-CathPCI) risk scores, we developed logistic regression and gradient descent boosting (XGBoost) models for each outcome using data from a prospective, all-comer, multicenter registry that enrolled consecutive coronary artery disease patients undergoing PCI in Japan between 2008 and 2020. The NCDR-CathPCI risk scores demonstrated good discrimination for each outcome (C-statistics of 0.82, 0.76, and 0.95 for AKI, bleeding, and in-hospital mortality) with considerable calibration. Compared with the NCDR-CathPCI risk scores, the XGBoost models modestly improved discrimination for AKI and bleeding (C-statistics of 0.84 in AKI, and 0.79 in bleeding) but not for in-hospital mortality (C-statistics of 0.96). The calibration plot demonstrated that the XGBoost model overestimated the risk for in-hospital mortality in low-risk patients. All of the original NCDR-CathPCI risk scores for adverse periprocedural events showed adequate discrimination and calibration within our cohort. When using the ML-based technique, however, the improvement in the overall risk prediction was minimal.

Leveraging Machine Learning Techniques to Forecast Chronic Total Occlusion before Coronary Angiography

“…It can learn quickly and efficiently from large amounts of data and its great flexibility makes it possible to learn even from missing data [ 20 ]. The XGBoost model had a much higher predictive accuracy compared to the generalized linear model, being able to capture complex associations in the data without requiring explicit high-order interactions and non-linear functions [ 21 ]. Using these features, predictive models could be developed from clinical demographics, echocardiography, and laboratory indexes, which are readily accessible and reproducible at admission.…”

Machine Learning Models for Predicting Acute Kidney Injury in Patients With Sepsis-Associated Acute Respiratory Distress Syndrome

“…Although the overfitting of training data is inevitable in some cases, the XGBoost model has been effectively implemented to solve many medical problems. For example, Niimi et al (30) developed logistic and XGBoost models to predict AKI in patients after percutaneous coronary intervention and found out that the XGBoost model was superior to the logistic regression model. Furthermore, Yue et al (31) introduced an XGBoost model to predict AKI in septic patients.…”

“…Yue et al (31) reported a machine learning model for predicting AKI in septic patients, but they only found that the status of mechanical ventilation was associated with a higher risk of AKI in sepsis patients. Another prediction model for predicting adverse events, including AKI after percutaneous coronary intervention (30), also missed detailed information about mechanical ventilation. In the present study, we reported the ventilation-associated variables (e.g., Pa o 2 /Fi o 2 ratio, PEEP), which were further confirmed to be associated with AKI.…”

Forewarned Is Forearmed: Machine Learning Algorithms for the Prediction of Catheter-Induced Coronary and Aortic Injuries

“…Both the regression and XGBoost were equally precise, with the regression model classifying more of the healthy persons correctly and XGBoost being more accurate in identifying individuals who later suffered a myocardial infarction; however, the XGBoost outperformed the logistic regression model according to the receiver operator characteristic scoring better in terms of accuracy in this metric. In contrast, Niimi et al compared NCDR-CathPCI risk scores for adverse periprocedural events (acute kidney injury, bleeding, and in-hospital mortality) with XGBoost models using data from a prospective, all-comer, multicentre registry from Japan, containing the records of 24,848 patients [ 8 ]. The XGBoost model modestly improved the discrimination for acute kidney injury and bleeding but not for in-hospital mortality, while it overestimated the risk for in-hospital mortality in low-risk patients.…”

“…The prediction of cardiovascular adverse events has been traditionally based on logistic regression modelling, itself a machine learning technique, with other, more advanced machine-learning algorithms gaining popularity only recently [ 7 , 8 ]. Several studies comparing conventional risk assessment methods with machine learning models in patients undergoing percutaneous coronary angioplasty reported a significantly improved performance and discrimination of the latter, while others showed only a modest improvement [ 8 , 9 , 10 , 11 ].…”