Diagnosis and classification of arrhythmia, which is associated with abnormal electrical activities in the heart, are critical for clinical treatments. Previous studies focused on the diagnosis of atrial fibrillation, which is the most common arrhythmia in adults. The classification performance achieved by studies on other arrhythmia types is not satisfactory for clinical use owing to the small number of classes (minority classes). In this study, we propose a novel framework for automatic classification that combines a residual network with a squeeze and excitation block, and a bidirectional long short-term memory. 8-class, 4-class, and 2-class performances were evaluated on the MIT-BIH arrhythmia database (MITDB), MIT-BIH atrial fibrillation database (AFDB), and PhysioNet/Computing in the cardiology challenge 2017 database (CinC DB), respectively, and they were superior to performance achieved by conventional methods. In addition, the class-wise F1-score in the minority classes was higher than those of the methods adopted in existing studies. To measure the generalization ability of the proposed framework, AFDB and CinC DB were tested using a MITDBtrained model, and superior performance was achieved compared with ShallowConvNet and DeepConvNet. We performed a crosssubject experiment using AFDB and obtained a statistically higher performance using the proposed method compared with typical machine learning methods. The proposed framework can enable the direct diagnosis of arrhythmia types in clinical trials based on the accurate detection of the minority class.

Background There is insufficient evidence for the use of single-lead electrocardiogram (ECG) monitoring with an adhesive patch-type device (APD) over an extended period compared to that of the 24-hour Holter test for atrial fibrillation (AF) detection. Objective In this paper, we aimed to compare AF detection by the 24-hour Holter test and 72-hour single-lead ECG monitoring using an APD among patients with AF. Methods This was a prospective, single-center cohort study. A total of 210 patients with AF with clinical indications for the Holter test at cardiology outpatient clinics were enrolled in the study. The study participants were equipped with both the Holter device and APD for the first 24 hours. Subsequently, only the APD continued ECG monitoring for an additional 48 hours. AF detection during the first 24 hours was compared between the two devices. The diagnostic benefits of extended monitoring using the APD were evaluated. Results A total of 200 patients (mean age 60 years; n=141, 70.5% male; and n=59, 29.5% female) completed 72-hour ECG monitoring with the APD. During the first 24 hours, both monitoring methods detected AF in the same 40/200 (20%) patients (including 20 patients each with paroxysmal and persistent AF). Compared to the 24-hour Holter test, the APD increased the AF detection rate by 1.5-fold (58/200; 29%) and 1.6-fold (64/200; 32%) with 48- and 72-hour monitoring, respectively. With the APD, the number of newly discovered patients with paroxysmal AF was 20/44 (45.5%), 18/44 (40.9%), and 6/44 (13.6%) at 24-, 48-, and 72-hour monitoring, respectively. Compared with 24-hour Holter monitoring, 72-hour monitoring with the APD increased the detection rate of paroxysmal AF by 2.2-fold (44/20). Conclusions Compared to the 24-hour Holter test, AF detection could be improved with 72-hour single-lead ECG monitoring with the APD.

BACKGROUND Cardiac arrest (CA) is the leading cause of death in critically ill patients, and clinical research has shown that the early identification of CA reduces mortality. Algorithms capable of predicting CA through multivariate time-series data with a high level of sensitivity have been developed. However, these algorithms suffer from a high level of false alarms, and their results are not clinically interpretable. OBJECTIVE Patients were retrospectively analyzed using data from Medical Information Mart for Intensive Care (MIMIC)-IV. Based on the multivariate vital signs of a 24-h time window for adults diagnosed with heart failure (HF), we extracted multi-resolution statistical features and cosine similarity-based features for the construction and development of gradient-boosting decision trees (DTs). Therefore, we propose cost-sensitive learning as a solution to this problem. Then, 10-fold cross-validation was performed to check the consistency of the model performance, and the Shapley additive explanations (SHAP) algorithm was used to capture the overall interpretability of the proposed model. METHODS From the Medical Information Mart for Intensive Care-IV, a total of patients were retrospectively analyzed. Based on multivariate vital signs of 24 h time window of adults with a diagnosis related to heart failure, we extracted multi-resolution statistical features and cosine similarity-based features for the construction and development of gradient-boosting decision trees. We proposed cost-sensitive learning as a solution to the imbalance problem. The 10-fold cross-validation was carried out to check the consistency of the model performance. To capture the overall interpretability of the proposed model, Shapley additive explanations value was used. RESULTS The proposed method yielded an overall area under the receiver operating characteristic curve (AUROC) of 0.86 and an area under the precision-recall curve (AUPRC) of 0.56. Regarding the early CA prediction performance of the proposed model, it achieved an AUROC above 0.8 in predicting CA events up to 6 h in advance. This result indicated that the prediction performance of the model was superior to those obtained in previous studies. Additionally, we demonstrated the effect of feature importance on the clinical interpretability of the proposed method and inferred differences in importance between the non-CA and CA groups. CONCLUSIONS The proposed framework can provide clinicians with more accurate CA prediction results for patients diagnosed with HF. Additionally, through its clinically interpretable prediction results, it can facilitate the understanding of clinicians. Furthermore, the similarity in vital sign changes can provide insights into understanding the temporal pattern changes in CA prediction for patients with HF-related diagnoses. CLINICALTRIAL N/A

BACKGROUND Wearable monitoring devices of an electrocardiogram (ECG) are widely used. However, the date of an adhesive single-lead ECG patch (SEP) in the diagnosis and 24-hour burden assessment of premature ventricular complex (PVC) is limited. OBJECTIVE We validated the diagnostic yield of SEP (mobiCARE™ MC-100, Seers Technology) for PVC detection and evaluated the variation of PVC burden monitored during three days. METHODS Patients with documented PVC on a 12-lead ECG were recruited. They underwent simultaneous ECG monitoring by the Holter and SEP on the first day. On the subsequent second and third days, ECG monitoring by SEP was continued and completed 3-day extended monitoring. We compared the detection yield of PVC on the first day of SEP to that of the Holter. Daily and 6-hour PVC burden variation was evaluated from the SEP data. The number of patients additionally identified to reach PVC thresholds of 10%, 15%, and 20% on 3-day extended monitoring of SEP and clinical factors associated with the higher burden variation were explored. RESULTS Recruited data of 134 monitored patients (mean age, 54.6 years; male, 33.6%) were analyzed. The median daily PVC burden was 2.4 (0.2-10.9) % by the Holter and 3.3 (0.3-11.7) % for the 3-day monitoring of SEP. The daily PVC burden detected on the first day of SEP was in agreement with that of the Holter: the mean difference was -0.07%, with 95% limits of agreement (-1.44%, 1.30%). A higher PVC burden on the first day was correlated with a higher daily (R2=0.34) and 6-hour burden variation (R2=0.48). Three-day monitoring by SEP identified 12/42 (28.6%), 10/56 (17.9%), and 4/60 (6.7%) more patients reaching 10%, 15%, and 20% of daily PVC burden, respectively. Younger age was associated with an additional detection of PVC burden with clinical significance on the extended monitoring (P=.02). CONCLUSIONS A SEP accurately detects PVC with a comparable diagnostic yield to the Holter. Three-day monitoring of PVC using SEP could be a practical alternative to identify more patients who cross the significant burden threshold. Young patients might require extended monitoring to determine the optimal treatment strategy for PVC.