JMIR Publications

ABSTRACT

Background:

Stress echocardiography (SE) is a well-established diagnostic tool in assessing patients with suspected coronary artery disease (CAD). Cardiovascular risk factors are used in the assessment of the probability of CAD. The link between the outcome of SE and patients’ variables including cardiovascular risk factors, current medication and anthropometric variables has not been widely investigated.

Objective:

This study aims to use Machine Learning (ML) to predict significant CAD defined by positive SE results in patients with chest pain based on patients’ anthropometrics, cardiovascular risk factors and medication as variables.

Methods:

A ML framework is proposed to automate the prediction of SE results. The proposed framework consists of four stages; feature extraction, pre-processing, feature selection and classification stage. A mutual information-based feature selection method was used to investigate the amount of information that each feature carries to define the positive outcome of SE. Two classification algorithms, Support Vector Machine (SVM) with Radial Basis Function (RBF) kernel, and Random Forest classifiers have been deployed. Data from 529 patients have been used to train and validate the proposed framework. Their mean age was 61 (±12 SD). The data consists of the anthropological data and cardiovascular risk factors such as gender, age, weight, family history, diabetes, smoking history, hypertension, hypercholesterolaemia, prior diagnosis of CAD and prescribed medications at the time of the test. The results of the SE were defined as outcome. A total of 82 patients had positive (abnormal) and 447 negative (normal) results, respectively. The proposed framework has been evaluated using the whole dataset including the cases with prior diagnosis of CAD. Five folds cross validation was used to validate the performance of the proposed framework. We also investigated the model in the subset of patients with no prior CAD.

Results:

The feature selection methods showed that prior diagnosis of CAD, sex, and prescribed medications such as angiotensin convertase enzyme inhibitor or angiotensin receptor blocker were the features that shared the most information about the outcome of SE. SVM classifiers showed the best trade-off between sensitivity and specificity and was achieved with three features. The best trade-off between sensitivity and specificity for the whole dataset accuracy was 66.63% with sensitivity and specificity 72.87%, and 67.67% respectively. However, for patients with no prior diagnosis of CAD only two features (sex and angiotensin convertase enzyme inhibitor or angiotensin receptor blocker use) were needed to achieve accuracy of 70.32% with sensitivity and specificity at 70.24%.

Conclusions:

This pilot study shows that ML can predict the outcome of SE in detecting significant CAD based on only a few features: patient prior cardiac history, gender, and prescribed medication. Further research recruiting higher number of patients who underwent SE could further improve the performance of the proposed algorithm with the potential of facilitating patient’s selection for early treatment / intervention with avoiding un-necessary downstream testing.