Cardiovascular diseases (CVDs) are the leading cause of premature death and disability globally, leading to significant increases in healthcare costs and economic strains. Artificial intelligence (AI) is emerging as a crucial technology in this context, promising to have a significant impact on the management of CVDs. A wide range of methods can be used to develop effective models for medical applications, encompassing everything from predicting and diagnosing diseases to determining the most suitable treatment for individual patients. This literature review synthesizes findings from multiple studies that apply AI technologies such as machine learning algorithms and neural networks to electrocardiograms, echocardiography, coronary angiography, computed tomography, and cardiac magnetic resonance imaging. A narrative review of 127 articles identified 31 papers that were directly relevant to the research, encompassing a broad spectrum of AI applications in cardiology. These applications included AI models for ECG, echocardiography, coronary angiography, computed tomography, and cardiac MRI aimed at diagnosing various cardiovascular diseases such as coronary artery disease, hypertrophic cardiomyopathy, arrhythmias, pulmonary embolism, and valvulopathies. The papers also explored new methods for cardiovascular risk assessment, automated measurements, and optimizing treatment strategies, demonstrating the benefits of AI technologies in cardiology. In conclusion, the integration of artificial intelligence (AI) in cardiology promises substantial advancements in diagnosing and treating cardiovascular diseases.

Background: Cardiogenic shock (CS) is a life-threatening complication of ST-elevation myocardial infarction (STEMI) and remains the leading cause of in-hospital mortality, with rates ranging from 5 to 10% despite advances in reperfusion strategies. Early identification and timely intervention are critical for improving outcomes. This study investigates the utility of machine learning (ML) models for predicting the risk of CS during the early phases of care—prehospital, emergency department (ED), and cardiology-on-call—with a focus on accurate triage and prioritization for urgent angiography. Results: In the prehospital phase, the Extra Trees classifier demonstrated the highest overall performance. It achieved an accuracy (ACC) of 0.9062, precision of 0.9078, recall of 0.9062, F1-score of 0.9061, and Matthews correlation coefficient (MCC) of 0.8140, indicating both high predictive power and strong generalization. In the ED phase, the support vector machine model outperformed others with an ACC of 78.12%. During the cardiology-on-call phase, Random Forest showed the best performance with an ACC of 81.25% and consistent values across other metrics. Quadratic discriminant analysis showed consistent and generalizable performance across all early care stages. Key predictive features included the Killip class, ECG rhythm, creatinine, potassium, and markers of renal dysfunction—parameters readily available in routine emergency settings. The greatest clinical utility was observed in prehospital and ED phases, where ML models could support the early identification of critically ill patients and could prioritize coronary catheterization, especially important for centers with limited capacity for angiography. Conclusions: Machine learning-based predictive models offer a valuable tool for early risk stratification in STEMI patients at risk for cardiogenic shock. These findings support the implementation of ML-driven tools in early STEMI care pathways, potentially improving survival through faster and more accurate decision-making, especially in time-sensitive clinical environments.

Background: While early risk stratification in STEMI is essential, the threat of cardiogenic shock (CS) persists after revascularization due to reperfusion injury and evolving instability. However, risk prediction in later phases—after revascularization—is less explored, despite its importance in guiding intensive care decisions. This study evaluates machine learning (ML) models for dynamic risk assessment in interventional cardiology and cardiac intensive care unit (CICU) phases, where timely detection of deterioration can guide treatment escalation. Methods: We retrospectively analyzed clinical and procedural data from 158 patients diagnosed with STEMI complicated by cardiogenic shock, treated between 2019 and 2022 at the Cardiology Department of the University Emergency Hospital of Bucharest, Romania. Machine learning models—Random Forest (RF), and Quadratic Discriminant Analysis (QDA)—were developed and tested specifically for the interventional cardiology and CICU phases. Model performance was evaluated using area under the receiver operating characteristic curve (ROC-AUC), accuracy (ACC), sensitivity, specificity, and F1-score. Results: In the interventional phase, RF and QDA achieved the highest accuracy, both reaching 87.50%. In the CICU, RF and QDA demonstrate the best performance, reaching ACCs of 0.843. QDA maintained consistent performance across phases. Relevant predictors included reperfusion strategy, TIMI flow before percutaneous coronary intervention (PCI), Killip class, creatinine, and Creatine Kinase Index (CKI)—all parameters routinely assessed in STEMI patients. These models effectively identified patients at risk for post-reperfusion complications and hemodynamic decline, supporting decisions regarding extended monitoring and ICU-level care. Conclusions: Predictive models implemented in advanced STEMI phases can contribute to dynamic, phase-specific risk reassessment and optimize CICU resource allocation. These findings support the integration of ML-based tools into post-PCI workflows, enabling earlier detection of clinical decline and more efficient deployment of intensive care resources. When combined with earlier-stage models, the inclusion of interventional and CICU phases forms a dynamic, end-to-end risk assessment framework. With further refinement, this system could be implemented as a mobile application to support clinical decisions throughout the STEMI care continuum.

Artificial intelligence (AI) is increasingly being integrated into medical practice, offering promising tools to improve diagnostic accuracy and clinical efficiency. In the field of breast pathology, AI applications, particularly those based on deep learning (DL) and machine learning (ML), are emerging as decision-support tools in both diagnostic and prognostic workflows. This review provides a comprehensive overview of current AI-based approaches, with a focus on their clinical utility in tumor detection, histological classification, biomarker assessment, and prediction of treatment response. In addition to summarizing available AI platforms, the review critically examines their level of clinical validation, regulatory status, and integration into routine practice. Key challenges are also discussed. Overall, AI is expected to play an increasingly important role in supporting pathologists and advancing precision medicine in breast cancer management.

Background: Suicide is a leading cause of preventable death, yet machine learning (ML) analyses of forensic (medico-legal) suicide data are scarce and, to our knowledge, absent for Romania. Population-based forensic registries offer exhaustive, autopsy-confirmed coverage that is structurally distinct from clinical or civil death-registration data. We applied supervised and unsupervised ML to a complete regional medico-legal suicide registry to profile the method of death and to identify latent victim subgroups of preventive relevance. Methods: We analysed 395 consecutive suicide deaths (Galați and Brăila counties, ≈750,000 inhabitants; 2018–2024). Two supervised classifiers—L2-regularised logistic regression (LR) and random forest (RF, 200 trees)—were trained to discriminate hanging from other methods, using eleven sociodemographic and clinical predictors, and evaluated by 10-fold stratified cross-validation. Given severe class imbalance, the area under the ROC curve (AUC) was the primary metric. Model hyperparameters were fixed a priori, and no class-imbalance correction was applied; both decisions are pre-specified and justified in the Methods. Robustness was assessed by stratified non-parametric bootstrap confidence intervals for the odds ratios, a tipping-point sensitivity analysis for the undocumented clinical fields, and Ward-linkage hierarchical clustering as an independent partitioning check. Predictor importance was quantified by out-of-bag (OOB) permutation importance and Spearman correlations. Unsupervised structure was assessed by K-means clustering (k = 2–7), with the optimal solution selected by the average silhouette coefficient and the elbow (WCSS) criterion. Reporting followed TRIPOD+AI and STROBE. Results: The study population was predominantly male (87.1%) and rural (73.2%), with a mean age of 54.1 years; hanging accounted for 94.2% of deaths—far above the European average (≈50%). RF achieved AUC = 0.865 ± 0.181 and LR AUC = 0.847 ± 0.192, both within the “excellent” discrimination band; sensitivity was very high (0.995–0.997) and specificity was limited (0.233–0.367), an expected consequence of imbalance. Prior suicide attempts (OOB importance 0.959; Spearman ρ = −0.549, p < 0.001; OR = 0.496, 95% CI 0.25–0.72) and the presence of a suicide note (importance 0.575; ρ = −0.439, p < 0.001; OR = 0.549, 95% CI 0.34–0.76) were the dominant predictors. K-means identified two well-separated clusters (silhouette = 0.707), and the partition was reproduced exactly by Ward-linkage hierarchical clustering (adjusted Rand index = 1.000). Cluster 2 (n = 19; 4.8%) was a clinically distinct, younger subgroup (42.4 vs. 54.7 years) characterised by prior attempts (57.9% vs. 0%), suicide notes (68.4% vs. 0%), higher psychiatric comorbidity (52.6% vs. 30.9%) and lower hanging proportion (36.8% vs. 97.1%)—an exploratory, hypothesis-generating profile of recurrent suicidal behaviour with documented prior contact with the medical or medico-legal system. The principal findings were stable across all plausible degrees of clinical under-documentation in the tipping-point sensitivity analysis. Conclusions: ML applied to a complete forensic suicide registry reproduced known regional epidemiology and, beyond classical statistics, isolated an exploratory but clinically coherent high-risk subgroup of direct relevance to the audit of structured post-attempt follow-up. This is, to our knowledge, the first ML study of Romanian forensic suicide data and supports integrating ML into medico-legal research and into the regional targeting and audit of existing post-attempt follow-up provision.

Background: Interrupted time-series studies have not shown a consistent increase in suicide mortality during COVID-19, but civil mortality registries cannot ascertain SARS-CoV-2 infection individually. We evaluated pandemic-related change in suicide mortality and the feasibility of integrating infection ascertainment into medico-legal surveillance. Methods: We analysed 395 suicide deaths in Galați and Brăila counties, Romania (January 2018–December 2024), where mandatory forensic autopsy ensures exhaustive ascertainment. The pre-specified primary analysis was seasonally adjusted segmented Poisson regression in January 2018–December 2022 with an onset in March 2020; the period 2023–2024 was excluded as incompletely ascertained. Changepoint detection and unadjusted period comparisons were descriptive. Infection was classified as laboratory-confirmed (documented ante-mortem RT-PCR) or possible/probable (compatible autopsy findings); no post-mortem testing was performed. Results: The level change at onset was not significant (IRR 1.29, 95% CI 0.80–2.08; p = 0.302), nor was the slope change (IRR 1.01, 95% CI 0.98–1.04; p = 0.639); power was 80% only against IRR ≥ 1.98. Posterior probabilities were 0.85 for any increase and 0.035 for a doubling. The unadjusted comparison (6.00 vs. 4.62 deaths/month; IRR 1.30, 95% CI 1.04–1.63) was descriptive; no changepoint coincided with onset. Among 140 deaths in 2020–2021, 34 (24.3%) met ascertainment criteria, 21 (15.0%) laboratory-confirmed. Conclusions: The data exclude a doubling of suicide mortality but not a modest increase: a regional surveillance signal, not a causal finding. Infection ascertainment proved feasible within routine medico-legal investigation, yielding a reproducible framework for other registries, an Eastern European series for future meta-analyses, and a proof-of-concept for a standing forensic surveillance capacity adaptable to future epidemics.