An Interpretable Hand-Crafted Feature-Based Model for Atrial Fibrillation Detection

Abstract

Atrial Fibrillation (AF) is the most common type of cardiac arrhythmia. Early diagnosis of AF helps to improve therapy and prognosis. Machine Learning (ML) has been successfully applied to improve the effectiveness of Computer-Aided Diagnosis (CADx) systems for AF detection. Presenting an explanation for the decision made by an ML model is considerable from the cardiologists' point of view, which decreases the complexity of the ML model and can provide tangible information in their diagnosis. In this paper, a range of explanation techniques is applied to hand-crafted features based ML models for heart rhythm classification. We validate the impact of the techniques by applying feature selection and classification to the 2017 CinC/PhysioNet challenge dataset. The results show the effectiveness and efficiency of SHapley Additive exPlanations (SHAP) technique along with Random Forest (RF) for the classification of the Electrocardiogram (ECG) signals for AF detection with a mean F-score of 0.746 compared to 0.706 for a technique based on the same features based on a cascaded SVM approach. The study also highlights how this interpretable hand-crafted feature-based model can provide cardiologists with a more compact set of features and tangible information in their diagnosis.

Artificial intelligence in cardiovascular prevention: new ways will open new doors

“…Many studies involving machine learning algorithms to identify atrial fibrillation from single-lead or ambulatory ECGs have been performed 58–76 . As an extensive database of single-lead ECGs is easily accessible, many machine learning models have been created compared with 12-lead ECGs.…”

“…Many studies involving machine learning algorithms to identify atrial fibrillation from single-lead or ambulatory ECGs have been performed. [58][59][60][61][62][63][64][65][66][67][68][69][70][71][72][73][74][75][76] As an extensive database of single-lead ECGs is easily accessible, many machine learning models have been created compared with 12-lead ECGs. Overall, accuracy, sensitivity and specificity tended to be above 0.85.…”

Primer on Machine Learning in Electrophysiology

“…Random forest algorithms are a popular approach in cardiac EP and have been used in studies ranging from predicting left atrial appendage flow velocity to predicting 30-day mortality post ST-elevation myocardial infarction. [47][48][49][50][51][52][53][54][55] Random forest models were useful in these studies because a decision path to the correct prediction could be found in the majority of distinct trees. Random forest models have been shown to be particularly useful in providing interpretability due to each split being a defined threshold that offers insight.…”

“…55 Rouhi et al showed that through extracting custom features from ECGs, a random forest model was able to detect AF. 48 Pičulin et al were able to predict the progression of hypertrophic cardiomyopathy 10 years out using a random forest model. 52 51 Random forest was able to predict well on this data set because it can split the multimodal data many ways to inform its decision and incorporate the correlation between the different data types.…”

Application of Dense Neural Networks for Detection of Atrial Fibrillation and Ranking of Augmented ECG Feature Set

“…Furthermore, attempts to reveal the importance of the input feature map have been presented with explanatory focus to the decision mechanisms in different machine learning classifiers. For example, the feature importance is estimated by the Gini index in a random forest classifier [ 31 ]; ‘weight’ method for Extreme Gradient Boosting classifier [ 31 ]; the bagged decision tree ensemble [ 29 ]; Gain index for decision trees [ 37 ]; minimum redundancy, maximum relevance and minimum redundancy maximum relevance justified with a decision trees ensemble [ 32 ]; minimum redundancy maximum relevance, decision tree and SVM ranking methods [ 33 ]; and logistic regression, permutation testing, random forest and SHapley Additive exPlanations (SHAP) [ 50 ]. The latter has shown that each feature importance technique results in different feature rankings, depending on their characteristics and assumptions.…”