Automatic detection of atrial fibrillation (AF) by cardiac devices is increasingly common yet sub-optimally groups AF, flutter or tachycardia (AT) together as ‘high rate events’. This may delay or misdirect therapy. We hypothesized that deep learning (DL) can accurately classify AF from AT by revealing electrogram (EGM) signatures. We studied 86 patients in whom the diagnosis of AF or AT was established at electrophysiological study (25 female, 65 ± 11 years). Custom DL architectures were trained to identify AF using N = 29,340 unipolar and N = 23,760 bipolar EGM segments. We compared DL to traditional classifiers based on rate or regularity. We explained DL using computer models to assess the impact of controlled variations in shape, rate and timing on AF/AT classification in 246,067 EGMs reconstructed from clinical data. DL identified AF with AUC of 0.97 ± 0.04 (unipolar) and 0.92 ± 0.09 (bipolar). Rule-based classifiers misclassified ~10–12% of cases. DL classification was explained by regularity in EGM shape (13%) or timing (26%), and rate (60%; p < 0.001), and also by a set of unipolar EGM shapes that classified as AF independent of rate or regularity. Overall, the optimal AF ‘fingerprint’ comprised these specific EGM shapes, >15% timing variation, <0.48 correlation in beat-to-beat EGM shapes and CL < 190 ms (p < 0.001). Deep learning of intracardiac EGMs can identify AF or AT via signatures of rate, regularity in timing or shape, and specific EGM shapes. Future work should examine if these signatures differ between different clinical subpopulations with AF.
There is a clinical spectrum for atrial tachyarrhythmias wherein most patients with atrial tachycardia (AT) and some with atrial fibrillation (AF) respond to ablation, while others do not. It is undefined if this clinical spectrum has pathophysiological signatures. This study aims to test the hypothesis that the size of spatial regions showing repetitive synchronized electrogram (EGM) shapes over time reveals a spectrum from AT, to AF patients who respond acutely to ablation, to AF patients without acute response. We studied n = 160 patients (35% women, 65.0 ± 10.4 years) of whom (i) n = 75 had AF terminated by ablation propensity matched to (ii) n = 75 without AF termination and (iii) n = 10 with AT. All patients had mapping by 64-pole baskets to identify areas of repetitive activity (REACT) to correlate unipolar EGMs in shape over time. Synchronized regions (REACT) were largest in AT, smaller in AF termination, and smallest in non-termination cohorts (0.63 ± 0.15, 0.37 ± 0.22, and 0.22 ± 0.18, P < 0.001). Area under the curve for predicting AF termination in hold-out cohorts was 0.72 ± 0.03. Simulations showed that lower REACT represented greater variability in clinical EGM timing and shape. Unsupervised machine learning of REACT and extensive (50) clinical variables yielded four clusters of increasing risk for AF termination (P < 0.01, χ2), which were more predictive than clinical profiles alone (P < 0.001). The area of synchronized EGMs within the atrium reveals a spectrum of clinical response in atrial tachyarrhythmias. These fundamental EGM properties, which do not reflect any predetermined mechanism or mapping technology, predict outcome and offer a platform to compare mapping tools and mechanisms between AF patient groups.
Atrial fibrillation (AF) is the most common arrhythmia in USA with more than 2.3 million people affected annually. Catheter ablation procedure is a method for treatment of AF, which involves 3D electroanatomic mapping of the patient's left atrium (LA) by maneuvering a conventional multipolar diagnostic catheter (MPDC) along the LA endocardial surface after which pulmonary vein (PV) isolation is performed, thus eliminating the AF triggers originating from the PVs. However, it remains unclear how to effectively utilize the information provided by the MPDC to locate the AF-sustaining sites, known as sustained rotor-like activities (RotAs). In this study, we use computer modeling to investigate the variations in the characteristics of the MPDC electrograms, namely, total conduction delay (TCD) and average cycle length (CL), as the MPDC moves towards a RotA source. Subsequently, a study with a human subject was performed in order to verify the predictions of the simulation study. The conclusions from this study may be used to iteratively direct an MPDC towards RotA sources thus allowing the RotAs to be localized for customized and improved AF ablation.
1 January 2025 in Circulation: Arrhythmia and Electrophysiology
BACKGROUND: Large language models (LLMs) such as Chat Generative Pre-trained Transformer (ChatGPT) excel at interpreting unstructured data from public sources, yet are limited when responding to queries on private repositories, such as electronic health records (EHRs). We hypothesized that prompt engineering could enhance the accuracy of LLMs for interpreting EHR data without requiring domain knowledge, thus expanding their utility for patients and personalized diagnostics. METHODS: We designed and systematically tested prompt engineering techniques to improve the ability of LLMs to interpret EHRs for nuanced diagnostic questions, referenced to a panel of medical experts. In 490 full-text EHR notes from 125 patients with prior life-threatening heart rhythm disorders, we asked GPT-4-turbo to identify recurrent arrhythmias distinct from prior events and tested 220 563 queries. To provide context, results were compared with rule-based natural language processing and Bidirectional Encoder Representations from Transformer-based language models. Experiments were repeated for 2 additional LLMs. RESULTS: In an independent hold-out set of 389 notes, GPT-4-turbo had a balanced accuracy of 64.3%±4.7% out-of-the-box at baseline. This increased when asking GPT-4-turbo to provide a rationale for its answers, a structured data output, and in-context exemplars, to a balanced accuracy of 91.4%±3.8% ( P <0.05). This surpassed the traditional logic-based natural language processing and BERT-based models ( P <0.05). Results were consistent for GPT-3.5-turbo and Jurassic-2 LLMs. CONCLUSIONS: The use of prompt engineering strategies enables LLMs to identify clinical end points from EHRs with an accuracy that surpassed natural language processing and approximated experts, yet without the need for expert knowledge. These approaches could be applied to LLM queries for other domains, to facilitate automated analysis of nuanced data sets with high accuracy by nonexperts.
Atrial fibrillation (AF) is a major cause of heart failure and stroke. The early maintenance of sinus rhythm has been shown to reduce major cardiovascular endpoints, yet is difficult to achieve. For instance, it is unclear how discoveries at the genetic and cellular level can be used to tailor pharmacotherapy. For non-pharmacologic therapy, pulmonary vein isolation (PVI) remains the cornerstone of rhythm control, yet has suboptimal success. Improving these therapies will likely require a multifaceted approach that personalizes therapy based on mechanisms measured in individuals across biological scales. We review AF mechanisms from cell-to-organ-to-patient from this perspective of personalized medicine, linking them to potential clinical indices and biomarkers, and discuss how these data could influence therapy. We conclude by describing approaches to improve ablation, including the emergence of several mapping systems that are in use today.
BACKGROUND: Drug-induced QT prolongation after successful inpatient loading of class III antiarrhythmics may occur during routine outpatient care. Insertable cardiac monitors offer continuous signals but are limited by single-lead configuration. We hypothesized that a spatially aware deep learning system (3DRECON-QT) can reconstruct spatial information from a single lead vector to quantify QT/QTc and identify high-risk prolongation. METHODS: We developed 3DRECON-QT using a multitask encoder–decoder that ingests a 10-s single-lead signal, reconstructs 12 leads, and predicts QT/QTc. The model was developed using 12-lead ECGs with clinician-adjudicated QT/RR from a large health system and tested in an external center with different ECG hardware. Continuous monitoring performance was assessed in a public dofetilide‐loading data set with serial ECGs. In a real-world cohort of outpatients on dofetilide or sotalol presenting to the hospital or emergency room for any reason, rates of ventricular arrhythmias and QT prolongation were assessed. Device validation was tested in patients with insertable cardiac monitor recordings paired with clinical 12-lead ECGs. RESULTS: 3DRECON-QT classified prolonged QTc from single-lead signals with area under the receiver operating characteristics curve, 0.942 (mean absolute error, 17.5 ms) in the internal test set and 0.943 (mean absolute error, 21.1 ms) externally. During continuous dofetilide monitoring, predictions correlated with ground truth ( r , 0.851; mean absolute error, 17.8 ms; area under the receiver operating characteristics curve, 0.936 for prolonged QTc, 0.816 for ≥15% QTc rise). QTc prediction from true insertable cardiac monitor recordings showed r =0.824 and mean absolute error, 17.5 ms. In outpatients on class III antiarrhythmics (n=1676), 16.5% had high-risk QTc prolongation. Ventricular arrhythmia events were 3.97% versus 0.86% without prolongation (adjusted odds ratio, 4.24 [95% CI, 1.81–9.90]). 3DRECON-QT detected these events with area under the receiver operating characteristics curve 0.94 (F1 score, 0.60). CONCLUSIONS: A single-lead, deep-learning approach can achieve guideline-level measurement accuracy, enable continuous QTc surveillance from nonstandard ECG vectors, and identify clinically meaningful outpatient QTc prolongation associated with a >4-fold increase in serious ventricular arrhythmias. This strategy may enhance safety monitoring after class III antiarrhythmic initiation and support targeted intervention.