Key result
Tree-based machine learning using sample entropy yields higher AFib classification accuracy than non-tree models.
Why the study?
Atrial fibrillation episodes are sporadic and may not appear during clinical visits, creating a risk of misdiagnosis that can lead to life-threatening strokes.
Can machine-aided analysis utilizing entropy and volatility of RR-intervals accurately classify atrial fibrillation?
Observational (n=23)
Can machine-aided analysis utilizing entropy and volatility of RR-intervals accurately classify atrial fibrillation?
Machine learning models utilizing Sample Entropy and volatility of RR-intervals, particularly tree-based methods, can effectively detect atrial fibrillation, highlighting potential applications for wearable heart rate monitoring devices.
Supports entropy-ML for wearable AF screening; hypothesis-generating and requires prospective validation before practice change.
Atrial Fibrillation (A-FIB) is the most common heartbeat arrhythmia in the United States. It is characterized by an irregularity in electrical impulse reception, causing the atria to beat asynchronously from the ventricles. A-FIB episodes are sporadic, so symptoms may not appear during clinical visits, causing a risk of misdiagnosis which can lead to life-threatening strokes. This study seeks to utilize machine-aided analysis to detect A-FIB through the entropy and volatility of RR-intervals, the distance between two heartbeat-peaks, in electrocardiograms (ECG). Data were retrieved from the Massachusetts Institute of Technology and Beth-Israel Hospital (MIT-BIH) A-FIB Database. Twenty-three subjects were analyzed using subsections, and 45\% of the RR-Intervals among the subjects were classified as A-FIB. ECG recordings for each subject were separated into subsections, collections of 25 RR-Intervals, and analyzed using Leave One Person Out Cross Validation (LOPO-CV). Volatility, an important predictor in LOPO-CV, was used to measure the dispersion of the length of the RR-intervals in seconds. Shannon, Approximate, and Sample Entropy were also used to classify A-FIB. These entropy features measured the irregularity of RR-Intervals, where decreased regularity suggested increased entropy values. Hyperparameter tuning on all tree-based methods and some non-tree-based methods was conducted to improve accuracy. The results of this study indicated that Sample Entropy was a strong predictor of A-FIB. Tree-based methods such as Random Forests and Cat Boosting had higher accuracies than non-tree-based models. If implemented in wearable heart rate monitoring devices like smartwatches, these results can aid in the preliminary diagnosis of A-FIB outside clinical settings.
No takes yet. Share an insight, caveat, or question.
Luke Ni (2023) conducted an observational in Atrial Fibrillation (n=23). Tree-based machine learning models using entropy and volatility vs. Non-tree-based models was evaluated on Classification accuracy and sensitivity for Atrial Fibrillation. Sample Entropy was a strong predictor of atrial fibrillation, and tree-based machine learning methods such as Random Forests and Cat Boosting achieved higher classification accuracies than non-tree-based models.
Synapse has enriched 5 closely related papers on similar clinical questions. Consider them for comparative context: