KBKelly A. Brennan

Kelly A. Brennan

*This is an algorithmically generated author record based on publicly available information.

Current Organizations / Institutes

The organizations/institutes from the scholar's latest article


Kelly A. Brennan has a total of 66 co-authors. Here are the top co-authors by number of shared publications:


Latest Publications

  • Open Access

    Feasibility of Noninvasive Atrial and Ventricular Activation, Conduction-Velocity, and Site-of-Origin Mapping with Solid-State Magnetocardiography: A Preclinical Validation Study

    by Kelly A. Brennan, Sabyasachi Bandyopadhyay, Charles Sillett, Jennifer K. Lyons, Makoto Kameno, Yasushi Terazono, Prasanth Ganesan, Xichong Liu, Gentaro Ikeda, Hiroyuki Takashima, et al.

    Background: Characterizing cardiac activation by its site of origin, propagation, and conduction velocity underlies arrhythmia diagnosis and management, but invasive electrophysiology (EP) mapping requires vascular access, fluoroscopy, and sedation. Magnetocardiography (MCG) enables contactless mapping, and recent solid-state sensors remove the cost, cryogenic, and shielding barriers of legacy systems. We assessed the feasibility of a novel solid-state MCG system for noninvasive arrhythmia site-of-origin (SOO) localization and activation reconstruction, benchmarked against electrocardiographic imaging (ECGi). Methods: In nine swine implanted with right atrial and right ventricular pacing leads, we recorded MCG and ECGi simultaneously during atrial and ventricular pacing. Invasive epicardial contact EP mapping provided the activation-time reference and MRI-derived lead-tip location the SOO reference. Local activation time (LAT), conduction velocity (CV), and SOO were compared on a co-registered chamber mesh. SOO error was the Euclidean distance to the MRI lead tip; LAT and CV agreement with EP were quantified by Pearson r and compared using Wilcoxon signed-rank tests. Results: Across 17 datasets (8 atrial, 9 ventricular), median SOO error was lower for MCG than ECGi in the atrium (19.6 vs 31.2 mm; p=0.023) and ventricle (12.0 vs 26.1 mm; p=0.074). LAT agreement with EP was comparable between modalities and higher in the ventricle (MCG r=0.63; ECGi r=0.68) compared with the atrium (MCG r=0.40; ECGi r=0.53), each correlating with invasive EP mapping above chance. CV agreement was modest and numerically higher for MCG in the ventricle. Conclusions: Solid-state MCG was feasible for noninvasive site-of-origin localization and activation mapping, with accuracy comparable to ECGi, motivating larger prospective studies to define its clinical role in noninvasive mapping. Clinical Perspective: What Is Known: Invasive electrophysiology mapping is the reference standard for localizing the arrhythmia site of origin and characterizing cardiac activation, but it requires vascular access, fluoroscopy, and prolonged sedation within the electrophysiology laboratory. Existing noninvasive options have important limitations: electrocardiographic imaging depends on concomitant thoracic imaging, size-dependent vest fit, and torso-conductivity modeling, while legacy magnetocardiography has required costly cryogenic and magnetically shielded infrastructure. What the Study Adds: In a preclinical large-mammal model referenced to invasive electrophysiology and MRI, a novel contactless solid-state magnetocardiography system feasibly localized atrial and ventricular sites of origin and reconstructed activation without body-surface electrodes. Magnetocardiography localized the site of origin more accurately than ECGi in the atrium, and showed numerically lower localization error in the ventricle. Magnetocardiography displayed activation-time and conduction-velocity agreement with invasive electrophysiology comparable to electrocardiographic imaging in both chambers. Solid-state magnetocardiography may offer a scalable, noninvasive approach for preprocedural localization and complementary activation mapping, motivating larger prospective clinical validation.

  • Top Cited Papers

    Deep Learning–Based Continuous QT Monitoring to Identify High-Risk Prolongation Events After Class III Antiarrhythmic Initiation

    by Rayan A. Ansari, Sabyasachi Bandyopadhyay, Rishi K. Trivedi, Kelly A. Brennan, Xichong Liu, Prasanth Ganesan, J. Weston Hughes, Alexander C. Perino, Euan A. Ashley, Paul J. Wang, et al.

    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.


  • The impact of preoperative antiobesity medications on weight loss in adolescents undergoing metabolic and bariatric surgery – a COSMIC study

    by Justine O. Chinn, Mark Shacker, Kelly A. Brennan, Michael Kochis, Alyssa Stetson, Christa Bizimana, Josélio Rodrigues de Oliveira Filho, Matthew A. Hornick, Janey S.A. Pratt, Marwa Abu El Haija, et al.

  • Open Access

    Reducing diverse sources of noise in ventricular electrical signals using variational autoencoders

    by Samuel Ruipérez-Campillo, Alain Ryser, Thomas M. Sutter, Brototo Deb, Ruibin Feng, Prasanth Ganesan, Kelly A. Brennan, Albert J. Rogers, Maarten Z.H. Kolk, Fleur V.Y. Tjong, et al.


  • Comparing Phenotypes for Acute and Long-Term Response to Atrial Fibrillation Ablation Using Machine Learning

    by Prasanth Ganesan, Maxime Pedron, Ruibin Feng, Albert J. Rogers, Brototo Deb, Hui Ju Chang, Samuel Ruiperez-Campillo, Viren Srivastava, Kelly A. Brennan, Wayne R. Giles, et al.

    BACKGROUND: It is difficult to identify patients with atrial fibrillation (AF) most likely to respond to ablation. While any arrhythmia patient may recur after acutely successful ablation, AF is unusual in that patients may have long-term arrhythmia freedom despite a lack of acute success. We hypothesized that acute and chronic AF ablation outcomes may reflect distinct physiology and used machine learning of multimodal data to identify their phenotypes. METHODS: We studied 561 consecutive patients in the Stanford AF ablation registry (66±10 years, 28% women, 67% nonparoxysmal), from whom we extracted 72 data features of electrograms, electrocardiogram, cardiac structure, lifestyle, and clinical variables. We compared 6 machine learning models to predict acute and long-term end points after ablation and used Shapley explainability analysis to contrast phenotypes. We validated our results in an independent external population of n=77 patients with AF. RESULTS: The 1-year success rate was 69.5%, and the acute termination rate was 49.6%, which correlated poorly on a patient-by-patient basis (φ coefficient=0.08). The best model for acute termination (area under the curve=0.86, Random Forest) was more predictive than for long-term outcomes (area under the curve=0.67, logistic regression; P <0.001). Phenotypes for long-term success reflected clinical and lifestyle features, while phenotypes for AF termination reflected electrical features. The need for AF induction predicted both phenotypes. The external validation cohort showed similar results (area under the curve=0.81 and 0.64, respectively) with similar phenotypes. CONCLUSIONS: Long-term and acute responses to AF ablation reflect distinct clinical and electrical physiology, respectively. This de-linking of phenotypes raises the question of whether long-term success operates through factors such as attenuated AF progression. There remains an urgent need to develop procedural predictors of long-term AF ablation success.


  • Engineering of Generative Artificial Intelligence and Natural Language Processing Models to Accurately Identify Arrhythmia Recurrence

    by Ruibin Feng, Kelly A. Brennan, Zahra Azizi, Jatin Goyal, Brototo Deb, Hui Ju Chang, Prasanth Ganesan, Paul Clopton, Maxime Pedron, Samuel Ruipérez-Campillo, et al.

    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 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.


  • Open Access

    Defibrillation effectiveness and safety of the shock waveform used in a contemporary wearable cardioverter defibrillator: Results from animal and human studies

    by Marye J Gleva, Joseph Sullivan, Thomas C Crawford, Greg Walcott, Ulrika Birgersdotter-Green, Kelley R Branch, Rahul N Doshi, Kaisa Kivilaid, Kelly Brennan, Ron K Rowbotham, et al.

    Introduction: The wearable cardioverter defibrillator (WCD) is used to protect patients at risk for sudden cardiac arrest. We examined defibrillation efficacy and safety of a biphasic truncated exponential waveform designed for use in a contemporary WCD in three animal studies and a human study.Methods: Animal (swine) studies:#1: Efficacy comparison of a 170J BTE waveform (SHOCK A) to a 150J BTE waveform (SHOCK B) that approximates another commercially available waveform. Primary endpoint first shock success rate.#2: Efficacy comparison of the two waveforms at attenuated charge voltages in swine at three prespecified impedances. Primary endpoint first shock success rate.#3: Safety comparison of SHOCK A and SHOCK B in swine. Primary endpoint cardiac biomarker level changes baseline to 6 and 24 hours post-shock.Human Study: Efficacy comparison of SHOCK A to prespecified goal and safety evaluation. Primary endpoint cumulative first and second shock success rate. Safety endpoint adverse events.Results: Animal Studies#1: 120 VF episodes in six swine. First shock success rates for SHOCK A and SHOCK B were 100%; SHOCK A non-inferior to SHOCK B (entire 95% CI of rate difference above -10% margin, p < .001).#2: 2,160 VF episodes in thirty-six swine. Attenuated SHOCK A was non-inferior to attenuated SHOCK B at each impedance (entire 95% CI of rate difference above -10% margin, p < .001).#3: Ten swine, five shocked five times each with SHOCK A, five shocked five times each with SHOCK B. No significant difference in troponin I (p = 0.658) or creatine phosphokinase (p = 0.855) changes from baseline between SHOCK A and SHOCK B.Human Study: Thirteen patients, 100% VF conversion rate. Mild skin irritation from adhesive defibrillation pads in three patients.Conclusions: The BTE waveform effectively and safely terminated induced VF in swine and a small sample in humans.Trial registration: Human study clinical trial registration:URL:https://clinicaltrials.gov; Unique identifier:NCT04132466.