Peripartum cardiomyopathy, one of the most fatal conditions during delivery, results in heart failure secondary to left ventricular systolic dysfunction. Left ventricular dysfunction can result in abnormalities in electrocardiography. However, the usefulness of electrocardiography in the identification of peripartum cardiomyopathy in pregnant women remains unclear.This study aimed to evaluate the effectiveness of a 12-lead electrocardiography–based artificial intelligence/machine learning-based software as a medical device for screening peripartum cardiomyopathy.This retrospective cohort study included pregnant women who underwent transthoracic echocardiography between a month before and 5 months after delivery and underwent 12-lead electrocardiography within 30 days of echocardiography between December 2011 and May 2022 at Seoul National University Hospital. The performance of 12-lead electrocardiography–based artificial intelligence/machine learning analysis (AiTiALVSD software; version 1.00.00, which was developed to screen for left ventricular systolic dysfunction in the general population) was evaluated for the identification of peripartum cardiomyopathy. In addition, the performance of another artificial intelligence/machine learning algorithm using only 1-lead electrocardiography to detect left ventricular systolic dysfunction was evaluated in identifying peripartum cardiomyopathy. The results were obtained under a 95% confidence interval and considered significant when P<.05.Among the 14,557 women who delivered during the study period, 204 (1.4%) underwent transthoracic echocardiography a month before and 5 months after delivery. Among them, 12 (5.8%) were diagnosed with peripartum cardiomyopathy. The results showed that AiTiALVSD for 12-lead electrocardiography was highly effective in detecting peripartum cardiomyopathy, with an area under the receiver operating characteristic of 0.979 (95% confidence interval, 0.953–1.000), an area under the precision-recall curve of 0.715 (95% confidence interval, 0.499–0.951), a sensitivity of 0.917 (95% confidence interval, 0.760–1.000), a specificity of 0.927 (95% confidence interval, 0.890–0.964), a positive predictive value of 0.440 (95% confidence interval, 0.245–0.635), and a negative predictive value of 0.994 (95% confidence interval, 0.983–1.000). In addition, a 1-lead (lead I) artificial intelligence/machine learning algorithm showed excellent performance; the area under the receiver operating characteristic, area under the precision-recall curve, sensitivity, specificity, positive predictive value, and negative predictive value were 0.944 (95% confidence interval, 0.895–0.993), 0.520 (95% confidence interval, 0.319–0.801), 0.833 (95% confidence interval, 0.622–1.000), 0.880 (95% confidence interval, 0.834–0.926), 0.303 (95% confidence interval, 0.146–0.460), and 0.988 (95% confidence interval, 0.972–1.000), respectively.The 12-lead electrocardiography–based artificial intelligence/machine learning–based software as a medical device (AiTiALVSD) and 1-lead algorithm are noninvasive and effective ways of identifying cardiomyopathies occurring during the peripartum period, and they could potentially be used as highly sensitive screening tools for peripartum cardiomyopathy.

We study scaling convolutional neural networks (CNNs), specifically targeting Residual neural networks (ResNet), for analyzing electrocardiograms (ECGs). Although ECG signals are time-series data, CNN-based models have been shown to outperform other neural networks with different architectures in ECG analysis. However, most previous studies in ECG analysis have overlooked the importance of network scaling optimization, which significantly improves performance. We explored and demonstrated an efficient approach to scale ResNet by examining the effects of crucial parameters, including layer depth, the number of channels, and the convolution kernel size. Through extensive experiments, we found that a shallower network, a larger number of channels, and smaller kernel sizes result in better performance for ECG classifications. The optimal network scale might differ depending on the target task, but our findings provide insight into obtaining more efficient and accurate models with fewer computing resources or less time. In practice, we demonstrate that a narrower search space based on our findings leads to higher performance.

Question answering (QA) in the field of healthcare has received much attention due to significant advancements in natural language processing. However, existing healthcare QA datasets primarily focus on medical images, clinical notes, or structured electronic health record tables. This leaves the vast potential of combining electrocardiogram (ECG) data with these systems largely untapped. To address this gap, we present ECG-QA, the first QA dataset specifically designed for ECG analysis. The dataset comprises a total of 70 question templates that cover a wide range of clinically relevant ECG topics, each validated by an ECG expert to ensure their clinical utility. As a result, our dataset includes diverse ECG interpretation questions, including those that require a comparative analysis of two different ECGs. In addition, we have conducted numerous experiments to provide valuable insights for future research directions. We believe that ECG-QA will serve as a valuable resource for the development of intelligent QA systems capable of assisting clinicians in ECG interpretations.

Electrocardiogram (ECG) synthesis is the area of research focused on generating realistic synthetic ECG signals for medical use without concerns over annotation costs or clinical data privacy restrictions. Traditional ECG generation models consider a single ECG lead and utilize GAN-based generative models. These models can only generate single lead samples and require separate training for each diagnosis class. The diagnosis classes of ECGs are insufficient to capture the intricate differences between ECGs depending on various features (e.g. patient demographic details, co-existing diagnosis classes, etc.). To alleviate these challenges, we present a text-to-ECG task, in which textual inputs are used to produce ECG outputs. Then we propose Auto-TTE, an autoregressive generative model conditioned on clinical text reports to synthesize 12-lead ECGs, for the first time to our knowledge. We compare the performance of our model with other representative models in text-to-speech and text-to-image. Experimental results show the superiority of our model in various quantitative evaluations and qualitative analysis. Finally, we conduct a user study with three board-certified cardiologists to confirm the fidelity and semantic alignment of generated samples. our code will be available at https://github.com/TClife/text_to_ecg

Electrocardiogram (ECG) is the first-line modality for identifying acute myocardial infarction (AMI) in patients with chest pain. However, the ECG can even be normal in AMI patients, thereby delaying diagnosis and adversely affecting prognosis. We aim to develop a deep-learning algorithm for detecting AMI using 12-lead ECG (DAMI-ECG).This study included retrospective cohorts from 2 separate hospitals. We developed and validated DAMI-ECG and estimated the diagnostic performance in patients who visited the emergency department (ED) for chest pain. Furthermore, we compared the accuracy of DAMI-ECG through interpretations of an ECG machine and experienced cardiologists.A total of 227,912 ECGs from 114,600 patients were used to develop and validate DAMI-ECG, and 2,274 ECGs from 1,765 patients who visited the ED for chest pain were used to estimate the diagnostic performance in a clinical setting. Among development and validation datasets, the area under the receiver operating characteristic curve of the DAMI-ECG for detecting AMI was 0.927 (95% confidence interval, 0.908–0.945) and 0.914 (0.899–0.929) during internal and external validations. The diagnostic accuracy of DAMI-ECG for detecting AMI in patients with chest pain at the ED was 86.1% (83.6–88.6%), which outperformed that of experienced cardiologists (78.6% [76.8–80.1%]) using ECG alone.In conclusion, the diagnostic performance of DAMI-ECG is excellent in detecting AMI in patients visiting the ED for chest pain and superior to that of cardiologists. This algorithm can be used in the ED or pre-hospital setting for early AMI diagnosis.

Based on the development of AI (artificial intelligence) and big data, an emerging number of methods achieved outstanding performance in myocardial infarction (MI) diagnosis by an electrocardiogram (ECG). However, conventional interpretation methods have low reliability for detecting MI and are challenging to apply to 2 leads of wearable devices. To evaluate whether the novel method can facilitate MI diagnosis by only 2-lead ECG. We propose T2T (from 2-lead to 12-lead), a deep generative model that simulates a standard 12-lead ECG from an input of two asynchronous leads by generating ten leads. We used and selected 15,012 ECGs (9,527 normal, 5,485 samples with any MI) from the Physikalisch-Technische Bundesanstalt dataset. This dataset was split into stratified training, validation, and test sets with a ratio of 7:1:2. Three models using 2-lead, T2T, and 12-lead ECG were developed and validated to predict a diagnosis of MI, and their accuracy was compared. The area under the receiver operating characteristic curves of the 2-lead, T2T and 12-lead ECG were 0.937, 0.948, and 0.960, respectively. The generated signals have less noise (seen in aVR and aVL in 2a and V4, V5, and V6) and baseline wander. For the V1 and V5 leads generated by T2T, the differences in amplitude are 6.4% and 7.3%, respectively, and the missing positional errors are under 10 ms. Novel ECG T2T algorithm demonstrated favorable performance in detecting MI using 12-lead ECG. The model is comprised of style, mapping, generative, and discriminative networks. Each network is built with residual blocks.The green line denotes the original signal, while the pink line represents the signals generated by ECGT2T.

To evaluate and visualize how artificial intelligence (AI) recognizes electrocardiography (ECG) characteristics of each heart failure (HF) subtype. We developed a visually explainable and interpretable AI-ECG algorithm for various HF subtypes to meet this need. This study proposes the ShapeExplainer (SE) to interpret models. From 33,920 patients at two hospitals, we trained a convolutional neural network to identify patients with HF subtypes using ECG and clinical diagnosis. HF subtypes include heart failure with reduced ejection fraction (HFrEF), heart failure with mid-range ejection fraction (HFmrEF, and heart failure with preserved ejection fraction (HFpEF).To interprete each HF classifier model, SE, which converts non-HF group data to be recognized as each HF subtype by each classifier, was trained. The converted fake ECG shows what the change is for the classifier to recognize as each HF subtype. SE from the non-HF group of the external validation set shows the result by comparing the shape before and after transformation. When tested on an independent set of 6,784 patients, the network model yielded values for the area under the curves of HFrEF of 0.945, HFmrEF of 0.884, and HFpEF of 0.859, respectively. The results of each classifier were visualized by overlapping the original and newly generated ECGs. Comparing each subtype and its fake ECG, there was no clear trend of electrocardiographic features. However, fake ECG showed a tendency of qrs amplitude as it progressed to non-HF, HFpEF, HFmrEF, and HFrEF, especially in T wave or ST-change. The proposed fake ECG generating method is a new approach for interpreting the AI-ECG model, helping physicians understand how AI recognizes HF subtypes.

To assess whether electrocardiography (ECG)-based artificial intelligence (AI) algorithm developed to detect 1-year mortality in patients with HF with reduced ejection fraction (HFrEF). Data from two hospital visits (2016 Oct to 2021 Apr) was used. Participants with good quality baseline ECGs and HFrEF were included. We tested the hypothesis that the application of AI to the ECG could identify 1-yr mortality in HFrEF patients. A total of 16,228 (64% male, mean age of 58.5) were eligible. When tested on an independent set of 16,228 patients, the network model yielded values for the area under the curve, sensitivity, specificity, and accuracy of 0.83, 83.3%, and 64.2%, respectively. Those with a positive AI screen were at 2.6 times the risk (p<0.0001) of developing future mortality compared with those with a negative screen . We used a sensitivity map to visualize the ECG region used in the AI model to detect high 1-yr mortality in HFrEF. The map shows that the AI model focused on the QRS complex, particularly the R-wave, in most patients.

Atrial fibrillation (AF) is the most cause of cardioembolic source causing cryptogenic stroke. In these, anticoagulation therapy could reduce recurrence of stroke. However, paroxysmal AF would not be detected even by 24 hours Holter monitoring. Deep learning-based electrocardiogram (ECG) analysis models were recently developed to detect AF during sinus rhythm.We aimed to develop a deep learning algorithm (DLA) to detect AF during sinus rhythm and validate the model in patients with cryptogenic stroke who underwent implantable cardiac monitoring (ICM) to diagnose paroxysmal AF. This cohort study involved three hospitals (A, B, and C). We developed a DLA to detect AF using sinus rhythm 10 s 12-lead ECG. We included adult patients aged ≥18 years from hospital A and B. We used development data from AF adult patients who had at least one atrial fibrillation rhythm in the study period (Jan 2016 to Dec 2021) and non-AF patients who had no reference to AF in the ECG and electronic medical record. DLA was based on convolutional neural network (CNN) using 10 s 12-lead. For external validation, the ECGs from 217 patients (hospital C) with cryptogenic stroke who underwent ICM were analyzed by using the DLA for validating the accuracy in the real-world clinical situations. We included 10,605 AF adult patients and 50,522 non-AF patients as development data. During the internal validation, the area under the curve (AUC) of the final DLA based on CNN was 0.793 (95% Confidence interval 0.778–0.807). In external validation data from cryptogenic stroke patients, the mean ICM duration was 15.1 months, and AF >5 mins was detected in 32 patients (14.5%). The diagnostic accuracy of DLA was 0.793 to detect AF during sinus rhythm, and AUC was 0.824. The sensitivity, specificity, positive predictive value, and negative predictive value of the model were 0.844, 0.784, 0.403, and 0.967, respectively, which outperformed other conventional predictive methods based on clinical factors, such as CHARGE-AF, C2hest, and HATCH. In this study, DLA accurately detected paroxysmal AF using 12-leads normal sinus rhythm ECG in patients with cryptogenic stroke and outperformed the conventional models. The DLA could be used as a screening tool to identify the cause of stroke in the future.

To develop deep learning models for predicting Interoperative hypotension (IOH) using waveforms from arterial blood pressure (ABP), electrocardiogram (ECG), and electroencephalogram (EEG), and to determine whether combination ABP with EEG or CG improves model performance. Data were retrieved from VitalDB, a public data repository of vital signs taken during surgeries in 10 operating rooms at Seoul National University Hospital from January 6, 2005, to March 1, 2014. Retrospective data from 14,140 adult patients undergoing non-cardiac surgery with general anaesthesia were used. The predictive performances of models trained with different combinations of waveforms were evaluated and compared at time points at 3, 5, 10, 15 minutes before the event. The performance was calculated by area under the receiver operating characteristic (AUROC), area under the precision-recall curve (AUPRC), sensitivity and specificity. The model performance was better in the model using both ABP and EEG waveforms than in all other models at all time points (3, 5, 10, and 15 minutes before an event) Using high-fidelity ABP and EEG waveforms, the model predicted IOH with a AUROC and AUPRC of 0.935 [0.932 to 0.938] and 0.882 [0.876 to 0.887] at 5 minutes before an IOH event. The output of both ABP and EEG was more calibrated than that using other combinations or ABP alone. The results demonstrate that a predictive deep neural network can be trained using ABP, ECG, and EEG waveforms, and the combination of ABP and EEG improves model performance and calibration.