Search Results (180)

Background: Diabetic foot ulcers (DFUs) are a severe complication of diabetes and are characterized by impaired wound healing and a high amputation risk. Exosomes—which are nanovesicles carrying proteins, RNAs, and lipids—mediate intercellular communication in wound microenvironments, yet their biomarker potential in DFUs remains underexplored. Methods: We analyzed transcriptomic data from GSE134431 (13 DFU vs. 8 controls) as a training set and validated findings in GSE80178 (6 DFU vs. 3 controls). A sum of 7901 differentially expressed genes (DEGs) of DFUs were detected and intersected with 125 literature-curated exosome-related genes (ERGs) to yield 51 candidates. This was followed by GO/KEGG analyses and a PPI network construction. Support vector machine–recursive feature elimination (SVM-RFE) and the Boruta random forest algorithm distilled five biomarkers (DIS3L, EXOSC7, SDC1, STX11, SYT17). Expression trends were confirmed in both datasets. Analyses included nomogram construction, functional and correlation analyses, immune infiltration, GSEA, gene co-expression and regulatory network construction, drug prediction, molecular docking, and RT-qPCR validation in clinical samples. Results: A nomogram combining these markers achieved an acceptable calibration (Hosmer–Lemeshow p = 0.0718, MAE = 0.044). Immune cell infiltration (CIBERSORT) revealed associations between biomarker levels and NK cell and neutrophil subsets. Gene set enrichment analysis (GSEA) implicated IL-17 signaling, proteasome function, and microbial infection pathways. A GeneMANIA network highlighted RNA processing and vesicle trafficking. Transcription factor and miRNA predictions uncovered regulatory circuits, and DGIdb-driven drug repurposing followed by molecular docking identified Indatuximab ravtansine and heparin as high-affinity SDC1 binders. Finally, RT-qPCR validation in clinical DFU tissues (n = 5) recapitulated the bioinformatic expression patterns. Conclusions: We present five exosome-associated genes as novel DFU biomarkers with diagnostic potential and mechanistic links to immune modulation and vesicular transport. These findings lay the groundwork for exosome-based diagnostics and therapeutic targeting in DFU management. Full article

The study of neuronal electrical activity and spatial organization is essential for uncovering the mechanisms that regulate neuronal electrophysiology and function. Mathematical models have been utilized to analyze the structural properties of neuronal networks, predict connectivity patterns, and examine how morphological changes impact neural network function. In this study, we aimed to explore the role of the actin cytoskeleton in neuronal signaling via primary cilia and to elucidate the role of the actin network in conjunction with neuronal electrical activity in shaping spatial neuronal formation and organization, as demonstrated by relevant mathematical models. Our proposed model is based on the polygamma function, a mathematical application of ramification, and a geometrical definition of the actin cytoskeleton via complex numbers, ring polynomials, homogeneous polynomials, characteristic polynomials, gradients, the Dirac delta function, the vector Laplacian, the Goldman equation, and the Lie bracket of vector fields. We were able to reflect the effects of neuronal electrical activity, as modeled by the Van der Pol equation in combination with the actin cytoskeleton, on neuronal morphology in a 2D model. In the next step, we converted the 2D model into a 3D model of neuronal electrical activity, known as a core-shell model, in which our generated membrane potential is compatible with the neuronal membrane potential (in millivolts, mV). The generated neurons can grow and develop like an organoid brain based on the developed mathematical equations. Furthermore, we mathematically introduced the signal transduction of primary cilia in neurons. Additionally, we proposed a geometrical model of the neuronal branching pattern, which we described as ramification, that could serve as an alternative mathematical explanation for the branching pattern emanating from the neuronal soma. In conclusion, we highlighted the relationship between the actin cytoskeleton and the signaling processes of primary cilia. We also developed a 3D model that integrates the geometric organization unique to neurons, which contains soma and branches, such that the mathematical model represents the interaction between the actin cytoskeleton and neuronal electrical activity in generating action potentials. Next, we could generalize the model into a cluster of neurons, similar to an organoid brain model. This mathematical framework offers promising applications in artificial intelligence and advancements in neural networks. Full article

This study introduces an innovative dog emotion classification system that integrates four non-invasive physiological indicators—skin potential (SP), muscle potential (MP), respiration frequency (RF), and voice pattern (VP)—with the extreme gradient boosting (XGBoost) algorithm. A four-breed dataset was meticulously constructed by recording and labeling physiological signals from dogs exposed to four fundamental emotional states: happiness, sadness, fear, and anger. Comprehensive feature extraction (time-domain, frequency-domain, nonlinearity) was conducted for each signal modality, and inter-emotional variance was analyzed to establish discriminative patterns. Four machine learning algorithms—Neural Networks (NN), Support Vector Machines (SVM), Gradient Boosting Decision Trees (GBDT), and XGBoost—were trained and evaluated, with XGBoost achieving the highest classification accuracy of 90.54%. Notably, this is the first study to integrate a fusion of two complementary electrophysiological indicators—skin and muscle potentials—into a multi-modal dataset for canine emotion recognition. Further interpretability analysis using Shapley Additive exPlanations (SHAP) revealed skin potential and voice pattern features as the most contributive to model performance. The proposed system demonstrates high accuracy, efficiency, and portability, laying a robust groundwork for future advancements in cross-species affective computing and intelligent animal welfare technologies. Full article

The increasing demands for the reliability of modern industrial equipment and structures necessitate advanced techniques for design, monitoring, and analysis. This review article presents the latest research advancements in the application of machine learning techniques to vibration and acoustic signal analysis from 2015 to 2024. A total of 96 peer-reviewed scientific publications were examined, selected using a systematic Scopus-based search. The main research areas include processes such as modeling and design, health management, condition monitoring, non-destructive testing, damage detection, and diagnostics. In the context of these processes, a review of machine learning techniques was conducted, including convolutional neural networks (CNNs), recurrent neural networks (RNNs), long short-term memory (LSTM), autoencoders, support vector machines (SVMs), decision trees (DTs), nearest neighbor search (NNS), K-means clustering, and random forests. These techniques were applied across a wide range of engineering domains, including civil infrastructure, transportation systems, energy installations, and rotating machinery. Additionally, this article analyzes contributions from different countries, highlighting temporal and methodological trends in this field. The findings indicate a clear shift towards deep learning-based methods and multisensor data fusion, accompanied by increasing use of automatic feature extraction and interest in transfer learning, few-shot learning, and unsupervised approaches. This review aims to provide a comprehensive understanding of the current state and future directions of machine learning applications in vibration and acoustics, outlining the field’s evolution and identifying its key research challenges and innovation trajectories. Full article

This article introduces a microwave sensor tailored for skin hydration monitoring. The design enables wireless operation by separating the sensing component from the reader, making it ideal for wearable devices like wristbands. The sensor consists of a semi-lumped $LC$ resonator coupled to an inductive coil reader, where the capacitive part of the sensing tag is in contact with the skin. The variations in the skin hydration level alter the dielectric properties of the skin, which, in turn, modify the resonances of the $LC$ resonator. Experimental in vivo measurements confirmed the sensor’s ability to distinguish between four hydration conditions: wet skin, skin treated with moisturizer, untreated dry skin, and skin treated with Vaseline, by measuring the resonance frequencies of the sensor. Measurement of the input reflection coefficient ($S_{11}$) using a vector network analyzer (VNA) revealed distinct reflection poles and zeros for each condition, demonstrating the sensor’s effectiveness in detecting skin hydration levels. The sensing principle was analyzed using an equivalent circuit model and validated through measurements of a fabricated sensor prototype. The results confirm in vivo skin hydration monitoring by detecting frequency shifts in the reflection response within the 50–200 MHz range. The measurements and data analysis show less than $0.037\%$ error in transmission zero ($f_z$) together with less than $1.5\%$ error in transmission pole ($f_p$) while being used to detect skin hydration status on individual human subjects. The simplicity of the detection method, focusing on key frequency shifts, underscores the sensor’s potential as a practical and cost-effective solution for non-invasive skin hydration monitoring. This advancement holds significant potential for skincare and biomedical applications, enabling detection without complex signal processing. Full article

Multilayer composite materials often develop internal defects at varying depths due to manufacturing and environmental factors. Traditional planar scanning methods lack the ability to pinpoint defect locations in depth. This study proposes a terahertz time-domain spectroscopy (THz-TDS)-based defect detection method using continuous wavelet transform (CWT) to convert spectral signals into time-frequency images. These are analyzed by the ResNet18 model combined with a support vector machine (SVM) classifier. Comparative experiments with four classical deep learning models and three classifiers show that the Residual Network with 18 layers (ResNet18-SVM) approach achieves the highest accuracy of 98.56%, effectively identifying three types of defects. The results demonstrate the method’s strong feature extraction, depth resolution, and its potential for nondestructive evaluation of multilayer structures. Full article

Bacteriological water quality monitoring is of utmost importance for safeguarding public health against waterborne diseases. Traditional methods such as membrane filtration (MF), multiple tube fermentation (MTF), and enzyme-based assays are effective in detecting fecal contamination indicators, but their time-consuming nature and reliance on specialized equipment and personnel pose significant limitations. This paper introduces a novel, portable, and cost-effective UV-LED/RGB water quality sensor that overcomes these challenges. The system is composed of a multi-well self-loading microfluidic device for sample-preparation-free analysis, RGB sensors for data acquisition, UV-LEDs for excitation, and a portable incubation system. Commercially available defined substrate technology, most probable number (MPN) analysis, and machine learning (ML) are combined for the real-time monitoring of bacteria colony-forming units (CFU) in a water sample. Fluorescence signals from individual wells are captured by the RGB sensors and analyzed using Multilayer Perceptron Neural Network (MLPNN) and Support Vector Machine (SVM) algorithms, which can quickly determine if individual wells will be positive or negative by the end of a 24 h period. The novel combination of ML and MPN analysis was shown to predict in 30 min the bacterial concentration of a water sample with a minimum prediction accuracy of 84%. Full article

Integrated sensing and communication (ISAC) can improve the energy harvesting (EH) efficiency of simultaneous wireless information and power transfer (SWIPT)-assisted IoT networks by enabling precise energy harvest. However, the transmit power is increased in the hybrid system due to the fact that the sensing signals are required to be transferred in addition to the communication data. This paper aims to tackle this issue by formulating an optimization problem to minimize the transmit power of the base station (BS) under a nonlinear EH model, considering the coexistence of power-splitting users (PSUs) and time-switching users (TSUs), as well as the beamforming vector associated with PSUs and TSUs. A two-layer algorithm based on semi-definite relaxation is proposed to tackle the complexity issue of the non-convex optimization problem. The global optimality is theoretically analyzed, and the impact of each parameter on system performance is also discussed. Numerical results indicate that TSUs are more prone to saturation compared to PSUs under identical EH requirements. The minimal required transmit power under the nonlinear EH model is much lower than that under the linear EH model. Moreover, it is observed that the number of TSUs is the primary limiting factor for the minimization of transmit power, which can be effectively mitigated by the proposed algorithm. Full article

In this paper, a fault diagnosis method for permanent magnet synchronous motors is proposed, combining wavelet packet transform (WPT) energy feature extraction and a genetic algorithm (GA)-optimized back propagation (BP) neural network. Firstly, for the common types of motor faults (turn-to-turn short-circuit, phase-to-phase short-circuit, loss of magnetism, inverter open-circuit, rotor eccentricity), a corresponding motor fault model is established. The stator current signals during motor operation are analyzed using wavelet packet transform, and energy features are extracted from them as feature vectors for fault diagnosis. Then, a BP neural network is constructed, and a genetic algorithm is used to optimize its initial weights and thresholds, thereby improving the network’s classification accuracy. The results show that the GA-BP model outperforms the SSA-PNN diagnostic model in terms of fault classification accuracy. In particular, for the diagnosis of normal operation, inverter open-circuit, and demagnetization faults, the accuracy rate reaches 100%. This method demonstrates high diagnostic accuracy and practical application value. Full article

Perception and sensing of the surrounding environment are crucial for ensuring the safety of autonomous driving systems. A key issue is securing sensor reliability from sensors mounted on the vehicle and obtaining accurate raw data. Surface contamination in front of a sensor typically occurs due to adverse weather conditions or particulate matter on the road, which can degrade system reliability depending on sensor placement and surrounding bodywork geometry. Moreover, the moisture content of dust contaminants can cause surface adherence, making it more likely to persist on a vertical sensor surface compared to moisture only. In this work, a 76–81 GHz radar sensor, a 72–82 GHz automotive radome tester, a 60–90 GHz vector network analyzer system, and a 76–81 GHz radar target simulator setup were used in combination with a representative polypropylene plate that was purposefully contaminated with a varying range of water and ISO standard dust combinations; this was used to determine any signal attenuation and subsequent impact on target detection. The results show that the water content in dust contaminants significantly affects radar signal transmission and object detection performance, with higher water content levels causing increased signal attenuation, impacting detection capability across all tested scenarios. Full article

Epilepsy is a neurologic condition characterized by recurring seizures resulting from aberrant brain activity. It is crucial to promptly and precisely detect epileptic seizures to ensure efficient treatment. The gold standard electroencephalography (EEG) accurately records the brain’s electrical activity in real time. The intent of this study is to precisely detect epileptic episodes by leveraging machine learning and deep learning algorithms on EEG inputs. The proposed approach aims to evaluate the feasibility of developing a novel technique that utilizes the Hurst exponent to identify EEG signal properties that could be crucial for classification. The idea posits that the prolonged duration of EEG in epileptic patients and those who are not experiencing seizures can differentiate between the two groups. To achieve this, we analyzed the long-term memory characteristics of EEG by employing time-dependent Hurst analysis. Together, the Hurst exponent and the Daubechies 4 discrete wavelet transformation constitute the basis of this unique feature extraction. We utilize the ANOVA test and random forest regression as feature selection techniques. Our approach creates and evaluates support vector machine, random forest classifier, and long short-term memory network machine learning models to classify seizures using EEG inputs. The highlight of our research approach is that it examines the efficacy of the aforementioned models in classifying seizures utilizing single-channel EEG with minimally handcrafted features. The random forest classifier outperforms other options, with an accuracy of 97% and a sensitivity of 97.20%. Additionally, the proposed model’s capacity to generalize unobserved data is evaluated on the CHB-MIT scalp EEG database, showing remarkable outcomes. Since this framework is computationally efficient, it can be implemented on edge hardware. This strategy can redefine epilepsy diagnoses and hence provide individualized regimens and improve patient outcomes. Full article

Crosstalk between transmission lines, primarily caused by capacitive coupling, is a major challenge in high-frequency electronic systems, leading to signal integrity degradation. This study investigates the effectiveness of capacitors placed between ground planes in ungrounded coplanar waveguide (UCPW) transmission lines fabricated on FR4 circuit boards. A vector network analyzer (VNA) was used to measure near-end crosstalk (S₃₁) reduction, with improvements of up to −40 dB observed. Experiments were conducted on transmission lines of 100 mm and 200 mm lengths, demonstrating the impact of capacitor placement on mitigating interference. The results indicate that this method provides a scalable and practical approach to improving signal integrity in compact, high-density electronic designs. These findings contribute to a deeper understanding of crosstalk mitigation strategies, offering valuable insights for applications in high-speed communication and RF circuit design. This work systematically analyzes the role of capacitor placement in reducing crosstalk, addressing a critical gap in the literature and paving the way for future advancements in transmission line optimization. Full article

The localization of radio emitters is a fundamental task in reconnaissance systems, and it has become increasingly important with the evolution of mobile networks. The signal Doppler frequency (SDF) method, developed for dual-use applications, leverages Doppler frequency shifts (DFSs) in received signals to estimate the positions of radio transmitters. This paper proposes enhancements to the SDF method through advanced signal processing techniques, including dedicated filtering and a novel two-level overlapping approach, which significantly improve localization accuracy. The overlapping technique increases the number of DFS estimations per time unit by analyzing overlapping segments at both the signal sample level and within the DFS vector. Simulation studies using various filter types and overlapping parameters were conducted to evaluate the effectiveness of these enhancements in a dynamic scenario involving multiple stationary transmitters and a single moving receiver. The results demonstrate that the proposed approach minimizes localization errors. The application of low-pass filtering at the DFS vector level improves localization accuracy. In the study, three types of filters for different cutoff frequencies are considered. Each of the analyzed filters with an appropriately selected cutoff frequency provides a comparable reduction in localization error at the level of about 30%. The use of overlapping at the signal sample level with a factor of 10% allows for more than a twofold decrease in localization errors, while overlapping at the DFS vector provides an increase in the refresh rate of the position of localized objects. Comparative analysis with direct position determination techniques additionally showed high effectiveness of the SDF method, especially using data filtration and overlapping. The simulation studies carried out are of significant importance for the selection of the operating parameters of real localization sensors in unmanned aerial vehicle (UAV) equipment. Full article

To investigate the relationship between metal magnetic memory testing (MMMT) signals and stress concentration factors (SCFs), four-level sinusoidal constant-amplitude load tension–tension fatigue tests were carried out on 45CrNiMoVA steel specimens with different SCFs. The normal component of MMMT signals, H_p(y), was collected during the fatigue tests, and three characteristics were extracted and analyzed during the tests, including the peak-to-peak value of abnormal peaks (ΔH_p(y)), the slope coefficient of the fitting line of H_p(y) (K₁), and the slope coefficient of the fitting line of H_p(y) between abnormal peaks (K₂), and a back-propagation (BP) neural network was developed to differentiate the SCF of the specimens. The results showed that both fatigue load and fatigue cycle number influenced MMMT signals, and the characteristics remained stable as the fatigue cycle number increased for the same fatigue load but increased significantly as fatigue load increased. In addition, all the characteristics increased as the distance between the scan line and the center line increased, but none of them could be used to differentiate the SCF of the specimens. With properly selected input vector and hidden nodes, the established BP neural network can quantitatively recognize the SCF of specimens. Full article

Fault location in DC microgrids (DCMGs) is a critical challenge due to the system’s inherent complexities and the demand for high reliability in modern power systems. This study proposes an explainable artificial intelligence (XAI)-based quantum deep neural network (QDNN) framework to address fault localization challenges in DCMGs. First, voltage signals from the DCMG are collected and analyzed using high-order synchrosqueezing transform to detect traveling waves (TWs) and extract critical fault parameters such as time of arrival, magnitude, and polarity of the first and second TWs. These features are fed into the proposed QDNN model that integrates advanced learning techniques for accurate fault localization. The cumulative distance from the fault point to the bus connecting the DCMG to the power network is considered the output vector. The model uses a combination of deep learning and quantum computing techniques to extract features and improve accuracy. To ensure transparency, an XAI technique called Shapley additive explanations (SHAP) is applied, enabling system operators to identify critical fault features. The SHAP-based explainability framework plays a critical role in translating the model’s predictions into actionable insights, ensuring that the proposed solution is not only accurate but also practically implementable in real-world scenarios. The results demonstrate the QDNN framework’s superior accuracy in fault localization even in noisy environments and with high-resistance faults, independent of voltage levels and DCMG configurations, making it a robust solution for modern power systems. Full article