Chao, Chieh-Ju

Abstract Echocardiography is the most widely used cardiac imaging modality, yet artificial intelligence-enabled interpretation remains limited by the inability of existing models to integrate visual assessment, quantitative measurement, and clinical reasoning within a unified framework. Here we present EchoAtlas, the first autoregressive vision-language model developed for echocardiographic interpretation. Trained on over 12.9 million question-answer pairs derived from approximately 2 million echocardiogram videos, EchoAtlas achieves 0.966 accuracy on multiple-choice questions in our internal test set and establishes a new state-of-the-art on the public MIMIC-EchoQA benchmark (0.699 vs. 0.508 previously). EchoAtlas also provides accurate quantitative measurements, segment-level regional wall motion assessment, longitudinal comparison, and diagnostic reasoning across diverse question formats — capabilities not previously demonstrated in this domain. These results highlight the potential of autoregressive vision-language models as a foundation for interactive echocardiographic interpretation, representing an early step toward scalable, auditable artificial intelligence systems in cardiology practice.

Key Points Atrial fibrillation after kidney transplantation is common and is associated with poor prognosis. Artificial intelligence applied to 12-lead electrocardiograms pretransplant accurately predicts the risk of new-onset atrial fibrillation in our cohort. High-risk artificial intelligence electrocardiography score was independently associated with increased risk of mortality and allograft failure. Background Incident atrial fibrillation (AF) is common after kidney transplantation (KTx) and is associated with worse clinical outcomes. Artificial intelligence electrocardiography (AI-ECG) algorithms have demonstrated efficacy in predicting risk of new-onset AF in the general population; however, their prognostic value in KTx recipients is relatively unknown. Methods Retrospective analysis was conducted on KTx recipients without AF, with at least one pretransplant ECG between 2011 and 2021 across three tertiary centers in the United States (Mayo Clinic sites in MN, AZ, and FL). A previously validated AI-ECG algorithm estimated the probability of incident AF for each patient. Based on AI-ECG probabilities, patients were categorized into high-risk and low-risk groups, with the optimal AI-ECG score cutoff determined. The incidence of new-onset AF, allograft failure, and mortality was compared between groups. Results Overall, 6246 patients (age 53.5±13.8 years; 58.9% male) were included. Pretransplant AI-ECG probability of AF ≥5% was the optimal cutoff for high risk of incident AF (sensitivity 72%, specificity 62%). High-risk scores were associated with true new-onset AF at 30 days (adjusted hazard ratio [aHR], 2.89; 95% confidence intervals [CI], 2.05 to 4.09; P < 0.001), 3 years (aHR, 2.54; 95% CI, 1.99 to 3.26, P < 0.001), and 5 years post-transplant (aHR, 2.48; 95% CI, 1.99 to 3.09, P < 0.001). High-risk AI-ECG scores were also associated with increased mortality (aHR, 1.56; 95% CI, 1.30 to 1.88, P < 0.001) and overall allograft failure (aHR, 1.50; 95% CI, 1.30 to 1.75, P < 0.001) through the 5-year follow-up. Conclusions This pretransplant AI-ECG parameter identified patients at increased risk of new-onset AF post-KTx and provided prognostic utility. Overall, this easy to obtain tool allows for risk stratification of patients who may benefit from closer monitoring, targeted risk factor modification, and early intervention.

Abstract Objective To develop machine learning models using patient and migraine features that can predict treatment responses to commonly used migraine preventive medications. Background Currently, there is no accurate way to predict response to migraine preventive medications, and the standard trial‐and‐error approach is inefficient. Methods In this cohort study, we analyzed data from the Mayo Clinic Headache database prospectively collected from 2001 to December 2023. Adult patients with migraine completed questionnaires during their initial headache consultation to record detailed clinical features and then at each follow‐up to track preventive medication changes and monthly headache days. We included patients treated with at least one of the following migraine preventive medications: topiramate, beta‐blockers (propranolol, metoprolol, atenolol, nadolol, timolol), tricyclic antidepressants (amitriptyline, nortriptyline), verapamil, gabapentin, onabotulinumtoxinA, and calcitonin gene‐related peptide (CGRP) monoclonal antibodies (mAbs) (erenumab, fremanezumab, galcanezumab, eptinezumab). We pre‐trained a deep neural network, “TabNet,” using 145 variables, then employed TabNet‐embedded data to construct prediction models for each medication to predict binary outcomes (responder vs. non‐responder). A treatment responder was defined as having at least a 30% reduction in monthly headache days from baseline. All model performances were evaluated, and metrics were reported in the held‐out test set (train 85%, test 15%). SHapley Additive exPlanations (SHAP) were conducted to determine variable importance. Results Our final analysis included 4260 patients. The responder rate for each medication ranged from 28.7% to 34.9%, and the mean time to treatment outcome for each medication ranged from 151.3 to 209.5 days. The CGRP mAb prediction model achieved a high area under the receiver operating characteristics curve (AUC) of 0.825 (95% confidence interval [CI] 0.726, 0.920) and an accuracy of 0.80 (95% CI 0.70, 0.88). The AUCs of prediction models for beta‐blockers, tricyclic antidepressants, topiramate, verapamil, gabapentin, and onabotulinumtoxinA were: 0.664 (95% CI 0.579, 0.745), 0.611 (95% CI 0.562, 0.682), 0.605 (95% CI 0.520, 0.688), 0.673 (95% CI 0.569, 0.724), 0.628 (0.533, 0.661), and 0.581 (95% CI 0.550, 0.632), respectively. Baseline monthly headache days, age, body mass index (BMI), duration of migraine attacks, responses to previous medication trials, cranial autonomic symptoms, family history of headache, and migraine attack triggers were among the most important variables across all models. A variable could have different contributions; for example, lower BMI predicts responsiveness to CGRP mAbs and beta‐blockers, while higher BMI predicts responsiveness to onabotulinumtoxinA, topiramate, and gabapentin. Conclusion We developed an accurate prediction model for CGRP mAbs treatment response, leveraging detailed migraine features gathered from a headache questionnaire before starting treatment. Employing the same methods, the model performances for other medications were less impressive, though similar to the machine learning models reported in the literature for other diseases. This may be due to CGRP mAbs being migraine‐specific. Incorporating medical comorbidities, genomic, and imaging factors might enhance the model performance. We demonstrated that migraine characteristics are important in predicting treatment responses and identified the most crucial predictors for each of the seven types of preventive medications. Our results suggest that precision migraine treatment is feasible.

Background: Conduction abnormality post-transcatheter aortic valve implantation (TAVI) remains clinically significant and usually requires chronic pacing. The effect of right ventricular (RV) pacing post-TAVI on clinical outcomes warrants further studies. Methods: We identified 147 consecutive patients who required chronic RV pacing after a successful TAVI procedure and propensity-matched these patients according to the Society of Thoracic Surgeons (STS) risk score to a control group of patients that did not require RV pacing post-TAVI. We evaluated routine echocardiographic measurements and performed offline speckle-tracking strain analysis for the purpose of this study on transthoracic echocardiographic (TTE) images performed at 9 to 18 months post-TAVI. Results: The final study population comprised 294 patients (pacing group n = 147 and non-pacing group n = 147), with a mean age of 81 ± 7 years, 59% male; median follow-up was 354 days. There were more baseline conduction abnormalities in the pacing group compared to the non-pacing group (56.5% vs. 41.5%. p = 0.01). Eighty-eight patients (61.6%) in the pacing group required RV pacing due to atrioventricular (AV) conduction block post-TAVI. The mean RV pacing burden was 44% in the pacing group. Left ventricular ejection fraction (LVEF) was similar at follow-up in the pacing vs. non-pacing groups (57 ± 13.0%, 59 ± 11% p = 0.31); however, LV global longitudinal strain (−12.7 ± 3.5% vs. −18.8 ± 2.7%, p < 0.0001), LV apical strain (−12.9 ± 5.5% vs. 23.2 ± 9.2%, p < 0.0001), and mid-LV strain (−12.7 ± 4.6% vs. −18.7 ± 3.4%, p < 0.0001) were significantly worse in the pacing vs. non-pacing groups. Conclusions: Chronic RV pacing after the TAVI procedure is associated with subclinical LV systolic dysfunction within 1.5 years of follow-up.

Abstract Objective To develop a natural language processing (NLP) algorithm that can accurately extract headache frequency from free‐text clinical notes. Background Headache frequency, defined as the number of days with any headache in a month (or 4 weeks), remains a key parameter in the evaluation of treatment response to migraine preventive medications. However, due to the variations and inconsistencies in documentation by clinicians, significant challenges exist to accurately extract headache frequency from the electronic health record (EHR) by traditional NLP algorithms. Methods This was a retrospective cross‐sectional study with patients identified from two tertiary headache referral centers, Mayo Clinic Arizona and Mayo Clinic Rochester. All neurology consultation notes written by 15 specialized clinicians (11 headache specialists and 4 nurse practitioners) between 2012 and 2022 were extracted and 1915 notes were used for model fine‐tuning (90%) and testing (10%). We employed four different NLP frameworks: (1) ClinicalBERT (Bidirectional Encoder Representations from Transformers) regression model, (2) Generative Pre‐Trained Transformer‐2 (GPT‐2) Question Answering (QA) model zero‐shot, (3) GPT‐2 QA model few‐shot training fine‐tuned on clinical notes, and (4) GPT‐2 generative model few‐shot training fine‐tuned on clinical notes to generate the answer by considering the context of included text. Results The mean (standard deviation) headache frequency of our training and testing datasets were 13.4 (10.9) and 14.4 (11.2), respectively. The GPT‐2 generative model was the best‐performing model with an accuracy of 0.92 (0.91, 0.93, 95% confidence interval [CI]) and R 2 score of 0.89 (0.87, 0.90, 95% CI), and all GPT‐2–based models outperformed the ClinicalBERT model in terms of exact matching accuracy. Although the ClinicalBERT regression model had the lowest accuracy of 0.27 (0.26, 0.28), it demonstrated a high R 2 score of 0.88 (0.85, 0.89), suggesting the ClinicalBERT model can reasonably predict the headache frequency within a range of ≤ ± 3 days, and the R 2 score was higher than the GPT‐2 QA zero‐shot model or GPT‐2 QA model few‐shot training fine‐tuned model. Conclusion We developed a robust information extraction model based on a state‐of‐the‐art large language model, a GPT‐2 generative model that can extract headache frequency from EHR free‐text clinical notes with high accuracy and R 2 score. It overcame several challenges related to different ways clinicians document headache frequency that were not easily achieved by traditional NLP models. We also showed that GPT‐2–based frameworks outperformed ClinicalBERT in terms of accuracy in extracting headache frequency from clinical notes. To facilitate research in the field, we released the GPT‐2 generative model and inference code with open‐source license of community use in GitHub. Additional fine‐tuning of the algorithm might be required when applied to different health‐care systems for various clinical use cases.

Aims:Increased left ventricular (LV) wall thickness is frequently encountered in transthoracic echocardiography (TTE). While accurate and early diagnosis is clinically important, given the differences in available therapeutic options and prognosis, an extensive workup is often required to establish the diagnosis. We propose the first echo-based, automated deep learning model with a fusion architecture to facilitate the evaluation and diagnosis of increased left ventricular (LV) wall thickness. Methods and Results: Patients with an established diagnosis of increased LV wall thickness (hypertrophic cardiomyopathy (HCM), cardiac amyloidosis (CA), and hypertensive heart disease (HTN)/others) between 1/2015 and 11/2019 at Mayo Clinic Arizona were identified. The cohort was divided into 80%/10%/10% for training, validation, and testing sets, respectively. Six baseline TTE views were used to optimize a pre-trained InceptionResnetV2 model. Each model output was used to train a meta-learner under a fusion architecture. Model performance was assessed by multiclass area under the receiver operating characteristic curve (AUROC). A total of 586 patients were used for the final analysis (194 HCM, 201 CA, and 191 HTN/others). The mean age was 55.0 years, and 57.8% were male. Among the individual view-dependent models, the apical 4-chamber model had the best performance (AUROC: HCM: 0.94, CA: 0.73, and HTN/other: 0.87). The final fusion model outperformed all the view-dependent models (AUROC: HCM: 0.93, CA: 0.90, and HTN/other: 0.92). Conclusion: The echo-based InceptionResnetV2 fusion model can accurately classify the main etiologies of increased LV wall thickness and can facilitate the process of diagnosis and workup.

Abstract Objective To compare the artificial intelligence‐enabled electrocardiogram (AI‐ECG) atrial fibrillation (AF) prediction model output in patients with migraine with aura (MwA) and migraine without aura (MwoA). Background MwA is associated with an approximately twofold risk of ischemic stroke. Longitudinal cohort studies showed that patients with MwA have a higher incidence of developing AF compared to those with MwoA. The Mayo Clinic Cardiology team developed an AI‐ECG algorithm that calculates the probability of concurrent paroxysmal or impending AF in ECGs with normal sinus rhythm (NSR). Methods Adult patients with an MwA or MwoA diagnosis and at least one NSR ECG within the past 20 years at Mayo Clinic were identified. Patients with an ECG‐confirmed diagnosis of AF were excluded. For each patient, the ECG with the highest AF prediction model output was used as the index ECG. Comparisons between MwA and MwoA were conducted in the overall group (including men and women of all ages), women only, and men only in each age range (18 to <35, 35 to <55, 55 to <75, ≥75 years), and adjusted for age, sex, and six common vascular comorbidities that increase risk for AF. Results The final analysis of our cross‐sectional study included 40,002 patients (17,840 with MwA, 22,162 with MwoA). The mean (SD) age at the index ECG was 48.2 (16.0) years for MwA and 45.9 (15.0) years for MwoA ( p < 0.001). The AF prediction model output was significantly higher in the MwA group compared to MwoA (mean [SD] 7.3% [15.0%] vs. 5.6% [12.4%], mean difference [95% CI] 1.7% [1.5%, 2.0%], p < 0.001). After adjusting for vascular comorbidities, the difference between MwA and MwoA remained significant in the overall group (least square means of difference [95% CI] 0.7% [0.4%, 0.9%], p < 0.001), 18 to <35 (0.4% [0.1%, 0.7%], p = 0.022), and 35 to <55 (0.5% [0.2%, 0.8%], p < 0.001), women of all ages (0.6% [0.3%, 0.8%], p < 0.001), men of all ages (1.0% [0.4%, 1.6%], p = 0.002), women 35 to <55 (0.6% [0.3%, 0.9%], p < 0.001), and men 18 to <35 (1.2% [0.3%, 2.1%], p = 0.008). Conclusions Utilizing a novel AI‐ECG algorithm on a large group of patients, we demonstrated that patients with MwA have a significantly higher AF prediction model output, implying a higher probability of concurrent paroxysmal or impending AF, compared to MwoA in both women and men. Our results suggest that MwA is an independent risk factor for AF, especially in patients <55 years old, and that AF‐mediated cardioembolism may play a role in the migraine–stroke association for some patients.

AbstractThe use of 3D technology has significantly improved the diagnostic accuracy of echocardiography by overcoming the limitation of 2D echocardiography. Transillumination 3D image post processing technique enhances shadows and contrast of cardiac structures with a movable virtual light source improving further the clarity and detail provided by conventional 3D imaging. In this report, we present 3 cases, one of mobile atherosclerotic aortic root plaque, another of small thrombi on ICD lead, and a third case of bicuspid aortic valve perforation due to infective endocarditis in whom TEE 3D transillumination further improved the diagnostic quality of conventional 3D imaging and lead to accurate clinical diagnosis.