Background: Large language models (LLMs) are increasingly proposed as clinical decision support tools. However, their reliability in the emergency department (ED) triage remains insufficiently validated. This study aimed to evaluate the performance and limitations of multiple LLMs in triage using a large retrospective dataset. Methods: We conducted a retrospective analysis of 39,375 anonymized patient cases from the ED of AHEPA University General Hospital, Thessaloniki, Greece (June 2024–July 2025), extracted from the hospital’s electronic medical record system. All cases were triaged in real time according to the Emergency Severity Index (ESI) by 25 emergency physicians. In cases of uncertainty, a senior emergency physician was consulted. Seven LLMs (ChatGPT-5 Thinking, ChatGPT-5 Instant, Gemini 2.5, Qwen 3, Grok 4.0, Deep Seek v3.1, and Claude Sonnet 4) were evaluated against the physician-assigned ESI level (reference standard). Outcomes included triage score agreement (quadratic weighted kappa, κw), clinic referral accuracy and admission prediction. Subgroup analyses were performed by referral clinic and admission outcome. The study was conducted in accordance with TRIPOD-AI reporting guidelines. Results: Model performance varied substantially. DeepSeek and Claude Sonnet 4 achieved the highest agreement with physician-assigned ESI (κw ≈ 0.467; raw accuracy: 61.7%). In contrast, GPT-5 Instant performed poorly across all evaluation metrics (κw = 0.176; 95% CI: 0.167–0.186). Claude Sonnet 4 demonstrated the best performance in clinic referral (67.1%; κ = 0.619) and admission prediction (κw ≈ 0.46). Subgroup analyses indicated higher performance in pediatric cases and organ-specific complaints, such as ophthalmology (up to 81% accuracy). LLMs also showed tendencies toward over- or under-triage. Conclusions: Current LLMs demonstrate promising but inconsistent capability in triage. While selected models achieved moderate alignment with physician ESI decisions, none achieved strong agreement (κ > 0.80). LLMs are most suitable as supervised decision support tools, particularly in anatomically well-defined clinical scenarios, rather than as autonomous systems.

Background: Novel and accessible biomarkers may add to the existing risk stratification schemes in patients with acute coronary syndrome (ACS). The platelet-to-lymphocyte ratio (PLR) and glucose-to-lymphocyte ratio (GLR) have emerged as potential indicators of systemic inflammation and metabolic stress, both of which are pivotal in ACS pathophysiology. The aim of this study was to investigate the prognostic significance of the PLR and GLR in patients with ACS. Methods: We performed a retrospective cohort study of patients hospitalized with ACS between 2017 and 2023 at Hippokration Hospital of Thessaloniki, Greece. PLR and GLR were calculated from admission blood samples. The primary endpoint was all-cause mortality. Logistic and Cox regression models were used to investigate the associations of PLR and GLR with all-cause mortality. Receiver operating characteristic (ROC) analysis, Kaplan–Meier survival curves, and restricted cubic spline (RCS) modeling were also applied. Results: In total, 853 patients (median age: 65 years, 72.3% males) were included. Higher PLR and GLR were independently associated with increased risk of long-term mortality [adjusted Odds Ratio (aOR) for PLR: 1.007, 95% CI: 1.005–1.008; and for GLR: aOR = 1.006, 95% CI: 1.003–1.008]. The optimal cut-off values were 191.92 for PLR and 66.80 for GLR. Kaplan–Meier and Cox regression analyses confirmed significantly reduced survival in patients with GLR and PLR values exceeding these thresholds. RCS analysis revealed non-linear relationships, with mortality risk rising sharply at higher levels of both markers. PLR showed superior prognostic performance (AUC: 0.673, 95% CI: 0.614–0.723) compared to GLR (AUC: 0.602, 95% CI: 0.551–0.653). Conclusions: While PLR demonstrated greater predictive accuracy, both PLR and GLR were consistently associated with mortality and may provide complementary prognostic information. Incorporating those ratios into routine clinical assessment may improve risk stratification, particularly in resource-limited settings or for patients without traditional risk factors.

Atrial fibrillation (AF) is the most common sustained arrhythmia and a growing global health burden, yet conventional monitoring with Holter devices, event recorders and implantable loop recorders often fails to adequately capture recurrence. Rapid advances in digital health, wearable biosensors and artificial intelligence (AI) have transformed consumer smartwatches and wearables into potential clinical tools capable of continuous, real-world rhythm surveillance. This narrative review synthesizes contemporary evidence on smartwatch-based AF monitoring, spanning core technologies—photoplethysmography, single-lead electrocardiography and AI fusion algorithms—and validation studies across post-ablation follow-up. Compared with traditional modalities, smartwatch-based AF monitoring demonstrates improved detection of AF recurrence, enhanced characterization of AF burden, symptom–rhythm correlation, and greater patient engagement. At the same time, key limitations are critically examined, including motion artifacts, false-positive alerts, short recording windows, adherence dependence, digital literacy and access gaps, as well as unresolved issues around regulation, interoperability and data privacy. By integrating engineering advances with guideline-directed care pathways, smartwatch-based AF monitoring holds promise to complement, rather than immediately replace, established diagnostic tools and to enable more proactive, individualized AF management. Future work must focus on robust clinical validation, equitable implementation and clear regulatory frameworks to safely scale these technologies.

Background: The platelet-to-hemoglobin ratio (PHR) has emerged as a potential prognostic marker in various cardiovascular contexts, but its role in acute coronary syndrome (ACS), particularly among patients with diabetes mellitus (DM), remains unclear. Methods: In this retrospective cohort study, 843 ACS patients admitted to the 2nd Cardiology Department at Hippokration Hospital of Thessaloniki, Greece, between 2017 and 2023 were evaluated. PHR was calculated from admission complete blood counts. The primary endpoint was all-cause mortality during a median follow-up of 25 months. Multivariate logistic and Cox regression analyses, receiver operating characteristic (ROC) curves, Kaplan–Meier survival analyses, and restricted cubic spline (RCS) models were employed, with subgroup analyses by DM status. Results: Higher PHR was independently associated with increased mortality in the overall cohort (adjusted hazard ratio [aHR] 1.35, p < 0.001). This association showed stronger predictive value in DM patients, reflected in both a higher aHR (1.52 vs. 1.36 in non-DM patients, p < 0.001 and p = 0.018, respectively) and superior discriminative performance on ROC analysis (AUC 0.707 vs. 0.600 overall, p = 0.0006). Kaplan–Meier analysis confirmed poorer survival in high-PHR groups, especially in DM patients. RCS analysis revealed a J-shaped relationship, with risk increasing markedly beyond PHR values of 2.2. Conclusions: PHR is an independent predictor of long-term mortality in ACS, with greater prognostic significance in DM patients. Its simplicity, low cost, and availability from routine blood tests make it a promising tool for risk stratification in ACS.

Ischemic heart disease remains the leading cause of death despite substantial advances in diagnosis, revascularization therapies, and risk-factor control. Beta-adrenergic receptor blockers (Beta-Blockers, BBs), long used to control heart rate, blood pressure, and reduce arrhythmic risk, may also confer cardioprotection through mechanisms beyond hemodynamic unloading. This review integrates an extensive range of preclinical, translational, and clinical studies to present a comprehensive overview of the cardioprotective effects of BBs in the context of myocardial ischemia and reperfusion injury. Mechanistic domains include modulation of redox homeostasis, attenuation of inflammation and neutrophil activation, preservation of mitochondrial integrity and anti-apoptotic signaling, improvement of endothelial function, and stabilization of calcium handling. Third-generation compounds, carvedilol and nebivolol, demonstrate additional antioxidant and vasodilatory benefits compared with first- and second-generation agents; however, no consistent class-wide effect exists across most pathways. The evidence base remains fragmented, often derived from agent- or context-specific studies in heterogeneous populations, with uncertainty surrounding optimal timing of intervention. By bridging mechanistic understanding with clinical outcomes, this review highlights the importance of standardized assessment of BB effects, the development of personalized treatment approaches, and the pursuit of future research to address ongoing translational gaps.

Dietary sodium restriction is widely recommended in heart failure (HF) management; however, its benefits and risks remain a subject of ongoing debate. While moderate sodium reduction may improve symptoms and quality of life in selected patients, excessive restriction can trigger maladaptive neurohormonal activation, worsen renal function, and increase the risk of hyponatremia, malnutrition, and cachexia. Patient response is heterogeneous, influenced by clinical risk profile, salt sensitivity, comorbidities, and age, with some high-risk patients experiencing neutral or adverse outcomes. Additional challenges arise from hidden sodium in processed foods, medications, and meals, which complicate monitoring and adherence. Effective sodium management in HF therefore requires a nuanced, individualized approach that integrates risk stratification, dietary counseling, and public health measures targeting the food industry. Future research should refine patient selection criteria and establish optimal sodium targets to balance therapeutic efficacy with safety in real-world practice.