Predicting risk of complications following a drug eluting stent procedure: A SVM approach for imbalanced data

Abstract

Drug Eluting Stents (DES) have distinct advantages over other Percutaneous Coronary Intervention procedures, but have recently been associated with the development of serious complications after the procedure. There is a growing need for understanding the risk of these complications, which has led to the development of simple statistical models. In this work, we have developed a predictive model based on Support Vector Machines on a real world live dataset consisting of clinical variables of patients being treated at a cardiac care facility to predict the risk of complications at 12 months following a DES procedure. A significant challenge in this work, common to most clinical machine learning datasets, was imbalanced data, and our results showed the effectiveness of the Synthetic Minority Over-sampling Technique (SMOTE) to address this issue. The developed predictive model provided an accuracy of 94% with a 0.97 AUC (Area under ROC curve), indicating high potential to be used as a decision support for management of patients following a DES procedure in real-world cardiac care facilities.

Conformal predictions in multimedia pattern recognition

“…In this work, the CP framework has been used to achieve this purpose. in the patient data, our experiments illustrated the effectiveness of the Synthetic Minority Over-sampling Technique (SMOTE) [78] to obtain good performance with Testing Exchangeability: As mentioned in the previous section, the only assumption for the CP framework to provide valid results is that the data should be i.i.d. ; rather, the data should be exchangeable (a weaker assumption, as stated earlier) i.e.…”

“…This is formulated as a binary classification problem which predicts the onset of complications, or otherwise. More details of the dataset can also be found in [131].…”

“…In addition to the contributions based on the CP framework, there were other contributions that were made to address related problems in cardiac decision support More details of this work can be found in [133] and [134]. Table 4.1 [139].…”

“…In one of the related contributions of this work [78], we performed a study that showed the effectiveness of Synthetic Minority Oversample Technique (SMOTE) as a data processing method to handle such class imbalance. However, when the CP framework is applied to such a scenario, the calibration property may not be as meaningful, since all the errors may be concentrated in a specific class (or a localized subset of classes).…”

Dynamic class imbalance learning for incremental LPSVM

“…Undersampling, oversampling, re-sampling has the majority/minority class examples randomly removed/duplicated respectively until a particular class distribution ratio is met (Liu, Wu, & Zhou, 2006Weiss, 2004). The synthetic minority oversampling technique (SMOTE) (Chawla, Bowyer, Hall, & Kegelmeyer, 2002) is a powerful method that has shown a great deal of success in SVM related applications (Gouripeddi et al, 2009;Maciejewski & Stefanowski, 2011;Wang, 2008). In general, sampling methods are applicable to various type of SVM algorithms being used (Batuwita & Palade, 2010) for data modeling, thus they are potential choices for SVM class imbalance learning, but they are not able to completely solve the class imbalance problem.…”