Abductive reasoning as a basis to reproduce expert criteria in ECG atrial fibrillation identification. (29th August 2018)

• Record Type: Journal Article
• Authors: Teijeiro, T; García, C A; Castro, D; Félix, P
• Is Part Of: Physiological measurement. Volume 39:Number 8(2018:Aug.)
• Publication Date: 2018-08-29
• Subjects: abductive reasoning -- atrial fibrillation -- ECG processing -- arrhythmia detection -- 2017 PhysioNet Challenge; Physiology -- Measurement -- Periodicals; Patient monitoring -- Periodicals; 612
• DOI: 10.1088/1361-6579/aad7e4
• Languages: English
• ISSNs: 0967-3334

Abstract

Abstract: Objective : This work aims at providing a new method for the automatic detection of atrial fibrillation, other arrhythmia and noise on short single-lead ECG signals, emphasizing the importance of the interpretability of the classification results. Approach : A morphological and rhythm description of the cardiac behavior is obtained by a knowledge-based interpretation of the signal using the Construe abductive framework. Then, a set of meaningful features are extracted for each individual heartbeat and as a summary of the full record. The feature distributions can be used to elucidate the expert criteria underlying the labeling of the 2017 PhysioNet/CinC Challenge dataset, enabling a manual partial relabeling to improve the consistency of the training set. Finally, a tree gradient boosting model and a recurrent neural network are combined using the stacking technique to provide an answer on the basis of the feature values. Main results : The proposal was independently validated against the hidden dataset of the Challenge, achieving a combined F 1 score of 0.83 and tying for the first place in the official stage of the Challenge. This result was even improved in the follow-up stage to 0.85 with a significant simplification of the model, attaining the highest score so far reported on the hidden dataset. Significance : The obtained results demonstrate the potential of Construe to provide robust and valuable descriptions of temporal data, even with the presence of significant amounts of noise. Furthermore, the importance of consistent classification criteria in manually labeled training datasets is emphasized, and the fundamental advantages of knowledge-based approaches to formalize and validate those criteria are discussed.