Eisuke Amiya

Was not provided for this article.

Introduction: Vaccine-related myocarditis (VRM) has gained attention during the global roll-out of COVID-19 vaccines. COVID-19 VRM is usually non-fatal, but factors associated with fatal outcomes remain unclear. We aimed to characterize features associated with reported fatal outcomes using a pharmacovigilance database. Methods: We retrospectively analysed COVID-19 VRM reports in the Japanese Adverse Drug Event Report database from 2021 to 2024. Reports were stratified into death and survival groups based on recorded outcome information. Logistic regression analyses identified factors associated with reported fatal outcomes. Results: 31 022 reports listed COVID-19 vaccines as suspected drug; 1041 reports included myocarditis [reporting odds ratio (ROR) 31.3, 95% CI 29.0–33.8]. The median time from vaccination to myocarditis onset was 2 (1–4) days, with most events occurring within 7 days. Among reports with VRM, men accounted for the majority (77.8%) and 68.6% were ≤39 years old. Outcome data were available for 863 reports, and death was recorded in 97 reports [11.2% (95% CI, 9.3%–13.5%)]. Reported fatal outcomes were more frequent in women, patients aged ≥70 years, those with delayed onset (≥8 days after COVID-19 vaccination), and those with ≥1 comorbidities. Multivariable logistic regression identified age ≥70 years as independently associated with reported fatal outcomes (OR 8.67, 95% CI 4.41–17.3; P < .001). Conclusion: Older age was associated with higher odds of reported fatal outcomes among VRM reports in JADER. Female sex, delayed onset, and comorbidities were more common among reports with fatal outcomes. These exploratory findings require validation in clinically adjudicated studies.

BACKGROUND Patients with heart failure present with various symptoms, and comprehensively identifying and accurately documenting them in clinical registries or structured databases requires substantial manual effort. Locally deployed large language models (LLMs), which process data entirely within institutional infrastructure, have enabled automated extraction of structured information from unstructured clinical text. However, symptom-specific extraction performance and reasoning-trace analyses of the underlying decision-making process remain largely unexplored. OBJECTIVE We utilized multiple locally deployed LLMs for symptom extraction from cardiology discharge summaries, examining how extraction performance varied across individual symptoms and characterizing the inferential processes through reasoning traces. METHODS Ten Japanese-language clinical summaries of patients with advanced heart failure evaluated for heart transplantation listing were included to assess the feasibility of locally deployed LLMs for automated symptom extraction, yielding 80 reference-standard symptom-case labels (10 cases × 8 symptoms). Gemma3-27b and Qwen3.5-9b were run locally on a single GPU to extract the presence or absence of eight symptoms required for transplant listing in Japan. Two prompting conditions were evaluated: (1) extraction of all symptoms from the full summary, and (2) an additional instruction to disregard symptoms unrelated to the index hospitalization. Phase 1 evaluated four non-reasoning conditions, and phase 2 added chain-of-thought reasoning to each condition. Results are reported descriptively, without formal statistical testing. RESULTS Overall accuracy across the four no-reasoning conditions ranged from 87.5% to 91.3%, with Qwen3.5 showing greater sensitivity to prompt modification than Gemma3. In symptom-specific evaluation across all conditions, palpitations and fatigability showed the highest false-positive rates (25% and 38%, respectively). Review of the reasoning outputs showed that, in some cases, both models inferred palpitations from electrocardiographic findings or documented tachycardia, and fatigability from the underlying diagnosis of heart failure, despite the absence of explicit symptom documentation. CONCLUSIONS Symptom extraction performance varies according to the linguistic and clinical characteristics of individual symptoms. Reasoning-trace analysis further showed that LLMs could infer symptom presence from objective clinical findings rather than relying solely on explicitly documented patient complaints.

Aims: Non-ischaemic dilated cardiomyopathy (DCM) is frequently characterized by the presence of pathogenic germline variants, and genotype positivity predicts poor prognosis. Despite its importance, genetic testing remains underutilized in the current era. Therefore, we aimed to develop a deep learning model to predict genotype positivity using echocardiographic videos. Methods and results: We included patients who were diagnosed with DCM and had genetic testing at the University of Tokyo Hospital, Japan, consecutively from 2014 to 2022. The apical four-chamber views of echocardiographic videos were collected. First, we developed a deep learning model based on the EchoNet-Dynamic model, and the area under the curve (AUC) was computed. Second, we calculated the Madrid genotype score (clinical scoring system) for each case. Third, we developed a logistic regression model that combined the Madrid genotype score and the deep learning model. Finally, we compared the AUC of the combined model with that of the Madrid genotype score alone by DeLong’s test. Out of the 258 patients, 117 patients (45.3%) had genetic variants, and 141 (54.7%) did not. TTN (30.8%) was the most common genotype, followed by LMNA (18.8%). The deep learning model yielded an AUC of 0.64. The Madrid genotype score was well validated and achieved an AUC of 0.73. The combined model yielded an AUC of 0.76 with a significant improvement from the Madrid genotype score alone (P = 0.03). Conclusion: The deep learning model demonstrated modest discriminative ability to predict genotype positivity using echocardiographic videos. The accuracy of the clinical scoring system improved when combined with the deep learning model.

Cardiac rehabilitation (CR) improves exercise capacity, but frequent cardiopulmonary exercise testing (CPET) is impractical. The AESCULON mini enables non‐invasive hemodynamic monitoring, though its role in CR remains unclear. Eleven patients (6 myocardial infarction, 3 angina pectoris, 2 dilated cardiomyopathy) undergoing outpatient CR at the University of Tokyo Hospital were studied. Hemodynamics were measured using the AESCULON mini before and after 20 min of aerobic exercise at the anaerobic threshold. CPET and brain natriuretic peptide (BNP) were assessed within 2 weeks. Stroke volume, cardiac output, and cardiac index tended to increase, and thoracic fluid content (TFC) decreased post‐exercise. TFC before ( r = 0.767, p = 0.006) and after ( r = 0.711, p = 0.014) correlated with BNP. Changes in stroke volume and cardiac output correlated with peak VO ₂ , percent predicted peak VO ₂ , and ΔVO ₂ /ΔWR. Patients with increased cardiac output during exercise had higher peak VO ₂ and ΔVO ₂ /ΔWR. Non‐invasive hemodynamic data from the AESCULON mini correlated with BNP and exercise capacity, suggesting its usefulness for detecting heart failure progression and estimating exercise capacity in CR.

Background: In patients with hypertrophic cardiomyopathy (HCM), genetic testing is crucial for cascade screening and risk stratification. However, it remains limited by financial and logistical constraints, necessitating prioritization. The Mayo HCM Genotype Predictor Score, based on clinical and echocardiographic variables, estimates genotype positivity with acceptable performance. Although sarcomeric variants are also associated with electrophysiological abnormalities, ECG parameters were not incorporated into this model. This study aimed to enhance genotype prediction in HCM by integrating ECG parameters. Methods: We retrospectively analyzed 466 patients with HCM from a Japanese multicenter cohort. Genotype positivity was defined as harboring pathogenic/likely pathogenic variants in sarcomere‐encoding genes. Candidate ECG variables were selected via multivariable logistic regression with bootstrap aggregation. A point‐based novel score was developed and internally validated using cross‐validation. Model performance was assessed by the area under the receiver operating characteristic curve and Akaike’s information criterion. Results: Genotype‐positive patients (30.3%) more frequently exhibited atrial fibrillation, intraventricular conduction disturbance, lower prevalence of high voltage, and abnormal T‐wave inversion in precordial leads than genotype‐negative patients and thus were incorporated into the novel Mayo‐ECG score. This score stratified genotype positivity from 7.1% (score ≤−1) to 91.4% (score ≥4), and its discriminative performance (area under the receiver operating characteristic curve, 0.81 [95% CI, 0.77–0.85]) outperformed the Mayo score (area under the receiver operating characteristic curve, 0.76 [95% CI, 0.71–0.81]; P =0.005) with better overall model fit (Akaike’s information criterion: 439 versus 479). Internal validation yielded consistent results with good calibration. Conclusions: The Mayo‐ECG improves genotype prediction, outperforming the conventional model. Given its simplicity, this model has the potential to prioritize genetic testing in HCM.