Signals

An electrocardiogram (ECG) is a vital diagnostic tool that provides crucial insights into the heart rate, cardiac positioning, origin of electrical potentials, propagation of depolarization waves, and the identification of rhythm and conduction irregularities. Analysis of ECG is essential, especially during pregnancy, where monitoring fetal health is critical. Fetal electrocardiography (fECG) has emerged as a significant modality for evaluating the developmental status and well-being of the fetal heart throughout gestation, facilitating early detection of congenital heart diseases (CHDs) and other cardiac abnormalities. Typically, fECG signals are acquired non-invasively through electrodes placed on the maternal abdomen, which reduces risk and enhances user convenience. However, these signals are often contaminated via various sources, including maternal electrocardiogram (mECG), electromagnetic interference from power lines, baseline drift, motion artifacts, uterine contractions, and high-frequency noise. Such disturbances impair signal fidelity and threaten diagnostic accuracy. This scoping review adhering to PRISMA-ScR guidelines aims to highlight the methods for signal acquisition, existing databases for validation, and a range of algorithms proposed by researchers for improving the quality of fECG. A comprehensive examination of 157,000 uniquely identified publications from Google Scholar, PubMed, and Web of Science have resulted in the selection of 6210 records through a systematic screening of titles, abstracts, and keywords. Subsequently, 141 full-text articles were considered eligible for inclusion in this study (from 1950 to 2026). By critically evaluating established techniques in the current literature, a strategy is proposed for analyzing fECG and calculating heart rate variability (HRV) for identifying fetal heart-related abnormalities. Advances in these methodologies could significantly aid in the diagnosis of fetal heart diseases, assisting timely clinical interventions and prevention.

Myocardial infarction (MI) remains one of the most critical causes of death worldwide, demanding predictive models that balance accuracy with clinical interpretability. This study introduces an explainable artificial intelligence (XAI) framework that integrates least absolute shrinkage and selection operator (LASSO) regression for feature selection, logistic regression for prediction, and Shapley additive explanations (SHAP) for interpretability. Using a dataset of 918 patients and 12 signal-derived clinical variables, the model achieved an accuracy of 87.7%, a recall of 0.87, and an F1 score of 0.89, confirming its robust performance. The key risk factors identified were age, fasting blood sugar, ST depression, flat ST slope, and exercise-induced angina, while the maximum heart rate and upward ST slope served as protective factors. Comparative analyses showed that the SHAP and p-value methods largely aligned, consistently highlighting ST_Slope_Flat and ExerciseAngina_Y, though discrepancies emerged for ST_Slope_Up, which showed limited statistical significance but high SHAP contribution. By combining predictive strength with transparent interpretation, this study addresses the black-box limitations of conventional models and offers actionable insights for clinicians. The findings highlight the potential of signal-driven XAI approaches to improve early detection and patient-centered prevention of MI. Future work should validate these models on larger and more diverse datasets to enhance generalizability and clinical adoption.

The early detection of smoke signals due to wildfires is vital in containing the extent of loss and reducing response time, particularly in inaccessible or forested areas. For lightweight object detection, this study contrasts the YOLOv9-tiny, YOLOv10-nano, YOLOv11-nano, YOLOv12-nano, and YOLOv13-nano algorithms in determining wildfire smoke at extended ranges. We present a robustness- and generalization-checking five-fold cross-validation. This study is also the first of its kind to train and publicly benchmark YOLOv10-nano up to YOLOv13-nano on the given dataset. We investigate and compare the detection performance against the standard performance metrics of precision, recall, F1-score, and mAP50, as well as the performance metrics regarding computational efficiency, including the training and testing time. Our results offer practical implications regarding the trade-off between pre-processing methods and model architectures for smoke detection when applied in real time on ground-based cameras installed on mountains and other high-risk fire locations. The investigation presented in this work provides a model in which implementations of lightweight deep learning models for wildfire early-warning systems can be achieved.