The object of the study is the process of radio signal delay and direction estimation using digital spectral-correlation analysis enhanced by machine learning. This process is essential for high-accuracy direction finding in electromagnetic monitoring systems. The problem addressed is the low adaptability and insufficient accuracy of traditional direction finding methods under variable signal conditions, especially due to manual parameter selection and the computational complexity of correlation processing. The essence of the obtained results is a machine learning-based method for predicting radio signal parameters (delay and angle), which reduced the standard deviation of direction finding estimates to 0.08–0.026° and delay estimation error to 1.5–14.8 μs across a signal-to-noise ratio range of 9 to 37 dB. These results are supported by averaging over 1000 realizations using Monte Carlo simulation, confirming their stability under noise. Due to its distinctive features, the proposed solution addressed the problem by enabling automated selection of processing parameters through a trained neural network that adapts to nonlinear signal characteristics, minimizing the need for manual adjustment or exhaustive search. These results are explained by the model’s ability to identify hidden dependencies between signal parameters and processing outcomes, enabling adaptive behavior and reduced deviations. Although no computational complexity assessment is provided, prediction-based parameter estimation is expected to improve processing speed in future implementations. The results can be applied in real-time electromagnetic monitoring, radio surveillance, and defense applications, especially under limited computing resources or varying noise conditions

A method of correlation-interferometric direction finding has been improved, which effectively solves the problem of radio direction finding of radio emission sources under conditions of exposure to one or two masking interference. The problem was solved using the selection of an unmasked fragment of the spatial spectrum of the signal and the reconstruction of the missing samples of its signal group. As a result of the synthesis of the proposed method, estimates of signal samples were obtained as exact solutions to the proposed energy balance equations. The resulting solutions provide a significant increase in the signal-to-interference ratio and, accordingly, direction-finding accuracy without increasing the number of reception channels of the antenna array. As a result of the simulation, the dependences of the standard deviation of the bearing estimate on the signal-to-noise ratio in the presence of interference were built. Under the influence of one or two masking interferences and a signal-to-interference ratio of 0 dB, the use of the known direction-finding method without interference selection produces an anomalously large direction-finding error of more than 0.42 degrees, which is practically independent of the signal-to-noise ratio. The direction-finding method with selection of spectral signal samples masked by interference reduces the direction-finding error to 0.22 degrees when exposed to one interference and to 0.3 degrees when exposed to two interferences. This is due to the presence of power losses of the usable signal during the selection of its samples masked by interference. The proposed method of direction finding with reconstruction of signal samples provides a significant gain in accuracy by 3–30 times compared to the method of selection of masked samples in the range of changes in the signal-to-noise ratio (–20.5) dB. The direction-finding error of the proposed method decreases with increasing signal/noise according to a hyperbolic dependence. It is advisable to use the proposed direction-finding method when masking no more than two samples of the signal group

Respiratory sound analysis is becoming increasingly popular as a non-invasive method for detecting adventitious sounds and aiding in the diagnosis of respiratory diseases such as chronic obstructive pulmonary disease (COPD), asthma, and pneumonia. However, studies in this field differ significantly in datasets, recording devices, annotation techniques, preprocessing pipelines, model architectures, validation strategies, and evaluation metrics. This scoping review examines current research on artificial intelligence-based respiratory sound analysis, with a focus on datasets, acquisition and annotation practices, signal processing, feature representations, machine learning (ML) and deep learning (DL) methods, and evaluation protocols. The search identified 1056 database records, and 89 reports were included for the final evidence mapping. The reviewed datasets support event-, cycle-, recording-, and patient-level tasks and differ considerably in population, scale, acquisition hardware, annotation granularity, and label structure. The findings also show that acquisition and annotation are closely linked, creating potential device-, recording-site-, and label-related confounding. Methodologically, the literature can be summarized in four broad stages: handcrafted feature-based ML; deep spectrogram learning; representation learning and multimodality; and deployment-oriented, robustness-focused systems. Despite recent achievements, the field remains limited by small and imbalanced datasets, annotation uncertainty, device and population variability, inconsistent data splitting, and limited external validation. More reliable clinical use will require standardized acquisition and annotation, quality-controlled preprocessing, patient-independent evaluation, task-appropriate metrics, transparent reporting, and validation across independent devices, datasets, and clinical populations.

Introduction: the study examined the integration of augmented reality technologies with deep learning in sports exercise monitoring, emphasizing its importance for improving precision, accuracy of execution, and motivation in both professional and educational contexts. Objective: the objective of the research was to develop and evaluate a real-time system for monitoring and classifying sports exercises with personalized recommendations. the scope included assessing the system’s effectiveness in improving exercise accuracy, learning retention, and participant motivation during pedagogical experiments. Methodology: the system applied pose estimation, joint angle validation, and action classification using deep learning models integrated with augmented reality feedback. a controlled pedagogical experiment was carried out with two groups, one experimental group using the system and one control group with traditional instruction. data collection involved pre-tests, post-tests, delayed post-tests, and motivational questionnaires. Results: the experimental group showed significant improvements in exercise accuracy, surpassing the control group’s performance. after two weeks, retention of correct technique remained higher, and participants reported increased interest, enjoyment, and perceived usefulness of training with the proposed system. Discussion: the results were consistent with previous literature emphasizing the role of immersive and interactive technologies in improving motor learning and motivation. however, this research extended earlier findings by confirming that real-time augmented reality guidance enhances both technical performance and long-term retention of exercise skills. Conclusions: the study concluded that the proposed system represents an effective and innovative pedagogical tool for advancing sports training, and its application should be expanded to broader educational and professional environments.

This research introduces a novel convolutional neural network-bidirectional long short-term memory (CNN-BiLSTM) hybrid network for the automatic classification of heart arrhythmias using 12-lead electrocardiograms (ECGs). By merging the spatial feature extraction capabilities of CNNs with the temporal precision of BiLSTM networks, our approach sets a new standard in cardiac diagnostics. The proposed model was tested against the comprehensive CPSC2018 dataset, demonstrating superior performance with an accuracy of 90.67%, precision of 93.27%, recall of 96.35%, and an F-score of 94.78%, surpassing existing state-of-the-art methods. These results underscore the effectiveness of integrating spatial and temporal data analysis, offering a robust and reliable tool for medical practitioners. This study represents a significant advancement in automated ECG analysis, paving the way for improved diagnosis and treatment of heart diseases, and contributing to enhanced patient outcomes in cardiac care.