Using Minimum Redundancy Maximum Relevance Algorithm to Select Minimal Sets of Heart Rate Variability Parameters for Atrial Fibrillation Detection

Abstract

Heart rate is quite regular during sinus (normal) rhythm (SR) originating from the sinus node. In contrast, heart rate is usually irregular during atrial fibrillation (AF). Complete atrioventricular block with an escape rhythm, ventricular pacing, or ventricular tachycardia are the most common exceptions when heart rate may be regular in AF. Heart rate variability (HRV) is the variation in the duration of consecutive cardiac cycles (RR intervals). We investigated the utility of HRV parameters for automated detection of AF with machine learning (ML) classifiers. The minimum redundancy maximum relevance (MRMR) algorithm, one of the most effective algorithms for feature selection, helped select the HRV parameters (including five original), best suited for distinguishing AF from SR in a database of over 53,000 60 s separate electrocardiogram (ECG) segments cut from longer (up to 24 h) ECG recordings. HRV parameters entered the ML-based classifiers as features. Seven different, commonly used classifiers were trained with one to six HRV-based features with the highest scores resulting from the MRMR algorithm and tested using the 5-fold cross-validation and blindfold validation. The best ML classifier in the blindfold validation achieved an accuracy of 97.2% and diagnostic odds ratio of 1566. From all studied HRV features, the top three HRV parameters distinguishing AF from SR were: the percentage of successive RR intervals differing by at least 50 ms (pRR50), the ratio of standard deviations of points along and across the identity line of the Poincare plots, respectively (SD2/SD1), and coefficient of variation—standard deviation of RR intervals divided by their mean duration (CV). The proposed methodology and the presented results of the selection of HRV parameters have the potential to develop practical solutions and devices for automatic AF detection with minimal sets of simple HRV parameters. Using straightforward ML classifiers and the extremely small sets of simple HRV features, always with pRR50 included, the differentiation of AF from sinus rhythms in the 60 s ECGs is very effective.

A New Approach to Detecting Atrial Fibrillation Using Count Statistics of Relative Changes between Consecutive RR Intervals

“…We have previously used the same data, consisting of RR intervals from over 60,000 1-min ECG segments with either AF or SR, in our studies [ 2 , 3 ]. In [ 2 ], we selected various HRV indices for the machine learning algorithms, whereas in [ 3 ], we determined the optimal x for pRRx. Although using the same database for different studies has some limitations, it also has practical benefits.…”

“…Many various heart rate variability (HRV) parameters are used alone or in combinations to distinguish atrial fibrillation (AF) from sinus rhythm (SR) [ 1 , 2 , 3 ]. We and some other authors have shown that the percentage of successive RR interval differences equal to or greater than 50 ms (pRR50) outperforms other HRV parameters for this purpose [ 4 ].…”

“…We and some other authors have shown that the percentage of successive RR interval differences equal to or greater than 50 ms (pRR50) outperforms other HRV parameters for this purpose [ 4 ]. Furthermore, machine learning (ML)-aided AF detection using combinations of HRV parameters gives the best results when pRR50 is included in all diagnostic models [ 2 ].…”

Identifying Complex Emotions in Alexithymia Affected Adolescents Using Machine Learning Techniques

“…Due to the enormous dimensionality of the dataset utilized in the proposed system, we implemented feature selection using the feature ranking technique in GNU Octave. The maximum relevance minimum redundancy method (mRMR) [ 33 ] is a feature extraction and selection method and helps to reduce computing complexity and create models with the best generalization, and it tends to improve the classifier’s learning efficiency [ 34 ]. In binary classification using τ-tests for independent precedents, feature selection plays a vital role between the classes.…”

Statistical and Diagnostic Properties of pRRx Parameters in Atrial Fibrillation Detection

“…Due to different distributions of HRV-derived parameters in AF and SR [ 8 ], HRV has gained new interest in AF detection in ECG [ 9 , 10 , 11 ], as well as from wearable devices [ 12 , 13 ]. Several authors used feature selection methods to find the most relevant HRV parameters for AF detection [ 14 , 15 ]. Others incorporated HRV to predict future occurrence of AF [ 16 , 17 ].…”