Using routinely collected clinical data for the early detection of liver disease is often challenging owing to class imbalance, limited feature dimensionality, and nonlinear relationships among biochemical markers. This study evaluates hybrid machine learning models of liver disease classification on a set of 583 patient records with ten clinical-biochemical features (Indian Liver Patient Dataset, ILPD). We performed standard preprocessing such as missing-value handling, feature scaling, and categorical encoding. Logistic regression baselines, support vector machines, shallow neural networks, and hybrid ensemble models based on stacking were tested. Repeated stratified 10 × 5 cross-validation was used to guarantee reliable estimation given a limited sample size, and the results were reported with mean performance, variability, and 95
This paper proposes a memristor-based artificial synapsis-like CMOS neuromorphic circuit with achieving Pavlov’s associative memory learning and forgetting functions. It consists of four key functional modules and can adaptively identify danger/temptation signals and then trigger the corresponding behavioral responses. The overall neuromorphic system is compacted into a single integrated circuit (IC) utilizing compact two-stage operational amplifiers (OpAmps). Large-size output transistors are placed by multi-finger layouts to ensure sufficient driving capability while maintaining a minimized layout area, with the smaller circuit dimension of only 0.011 mm2 layout area, and the lower power consumption of only 37.38 mW. 180 nm/1.8 V standard BCD process is adopted to implement the circuit design and performance simulation on Cadence. The execution results demonstrate that, the weight value of the proposed memristor-based artificial synaptic can be accurately adjusted in training process, and further the whole human-like neuromorphic system can effectively imitate the Pavlov’s conditioned reflex such as associative memory learning and forgetting process.
A novel bifusiform thrombectomy device driven by the two-way shape memory effect (TWSME) was proposed for the therapeutic challenge of distal medium vessel occlusion (DMVO) in acute ischemic stroke (AIS). The retriever was constructed by intertwining NiTi shape memory alloy (SMA) wires into a bifusiform shape, and an analytical electro-thermal model of the SMA wire was established to predict its equilibrium temperature under different direct current (DC) powers. Moreover, finite element modeling and analysis were conducted for identification of expected temperature distribution across the SMA wire woven bifusiform retriever by electrical activation. Fabrication and experimental testing of the retriever prototype were completed, and an optimized annealing process of the raw SMA wire for the TWSME device was found: muffle furnace pretreatment at 500 °C for 180 min, followed by 5-cycle stretching at 125 MPa, primary limited aging at 350 °C for 30 min, and secondary unconstrained aging at 300 °C for 60 min. A physical cerebrovascular model and a blood clot made of glycerol and water were used for in vitro thrombectomy tests. The results showed that with a driving current of 0.22 A, the retriever achieved a contraction ratio of 38.89
The increasingly complex functionalities of smart grids pose challenges to the performance of power grid equipment. Compared to traditional circuit architectures, the System-on-Chip (SoC) has gained popularity in smart grids due to its high robustness and low cost. However, current applications of SoC in power grid devices primarily focus on reducing power consumption and optimizing communication mechanisms, often neglecting the multi-dimensional and real-time requirements of packet data processing. To achieve the multi-dimensional real-time processing of packet data, this paper proposes a multi-core SoC-based protection and control device. The ARM core handles the packet preprocessing and multi-task management, while the FPGA executes the parallel processing of multiple tasks. A task management IP core is designed to enhance the multi-task management efficiency. The experimental results show that the proposed protection control device can complete the configuration of a single task module in microseconds, and the time consumed for the parallel execution of multiple modules is much shorter than the linear superposition of the time consumed by individual execution, and the task with small memory dependence has a stronger time-consuming advantage when executed in parallel. It effectively meets the real-time requirements for the multi-dimensional processing of message information in smart grid equipment.
The rapid expansion of the Internet of Things (IoT) has intensified the need for energy-efficient and balanced communication in resource-constrained Wireless Sensor Networks (WSNs). This paper introduces Lotus Optimization-based Load Balancing for Energy-Efficient Routing in Cluster-Based IoT (LOLB-IoT) Networks which employs Lotus Effect Optimization Algorithm (LEOA) to enhance load distribution and prolong network lifetime in heterogeneous IoT-enabled WSNs. Unlike traditional clustering protocols, LOLB-IoT employs a multi-objective fitness function that accounts for cluster density, node residual energy, and distance to the base station for effective Cluster Head (CH) selection. Additionally, a dual-gateway mechanism dynamically assigns inner and outer circle gateway nodes based on proximity to the centrally located base station, enabling efficient data forwarding and reducing transmission overhead. Drawing inspiration from the adaptive and self-cleaning properties of lotus leaves, LEOA integrates global (pollination-inspired) and local (droplet-inspired) search strategies to ensure robust CH selection and adaptive routing. MATLAB simulations reveal that LOLB-IoT consistently surpasses benchmark protocols such as OCHL, SWARAM, PSOBS, and DCH-GA, achieving up to 186.1
Heart disease is a leading cause of mortality all over the world. Electrocardiogram (ECG) is the less expensive non-invasive procedure among the diverse diagnosis techniques. However, it also faces challenges like the complexity of ECG interpretations, scarcity of medical experts, heart disease comorbidity and heart disease’s manifestation similarities in ECG signals. Machine learning and Deep Learning algorithms are feasible alternatives to the traditional heart disease diagnosis methods. For that reason, we have proposed hybrid Improved LeNet with LinkNet model to detect the heart disease using ECG signal images, which includes 5 phases like pre-processing, feature extraction, feature selection, data augmentation and classification. Initially, the input ECG signal images were pre-processed or denoised using bilateral filtering technique. Subsequently, features like colour, hierarchy of skeleton, Improved Local Gradient Patterns, deep features and statistical features were extracted from the denoised images. Then, from these retrieved features, most relevant or suitable features were chosen using Hybrid Improved Filter and Wrapper-based feature selection technique. Then these chosen features subjected to augmentation using random sampling process. Finally, this augmented feature is subjected to classification phase, where a hybrid model (ILN-LNK), that is the combination of Improved LeNet (ILN) and LinkNet (LNK) is utilized for heart disease classification. The proposed method was evaluated on the Mendeley ECG Images Dataset of Cardiac Patients. Moreover, this ILN-LNK model obtained superior accuracy and sensitivity of 0.991 and 0.975 at 90
This study presents a compact ultra-wide band (UWB) multiple-input multiple-output (MIMO) antenna system that operates over the 3.9 to 13.4 GHz frequency band. A new decoupling structure has been developed that separates antenna elements to reduce mutual coupling between them; this allows for improved isolation across the entire operating band. The proposed system has achieved an isolation level of less than − 18 dB and particularly good isolation in the range of 8.0 to 13.4 GHz. A second evaluation of the overall performance of the antenna system includes an evaluation with respect to MIMO diversity parameters such as envelope correlation coefficient (ECC), diversity gain (DG), radiation patterns and realized gain; all of which indicate stable performance over the entire UWB band. A prototype of the proposed antenna was fabricated and tested and found to agree closely to simulated performance. In addition, supervised machine learning regression models can be used to estimate certain antenna characteristics such as return loss and voltage standing wave ratio (VSWR) from simulated data. The accuracy of the predictions made by these models was verified using several statistical metrics including R-squared (R²), root mean square error (RMSE), mean square error (MSE) and mean absolute error (MAE). Of all the regression models that were evaluated, the R²-based regression model demonstrated the most predictive accuracy at approximately 98
The evolution of Medical telemetry and wireless body area networks (WBANs) systems demands highly efficient, compact, and intelligent antenna systems capable of maintaining reliable communication across dynamic environments. Traditional MIMO antenna arrays face challenges in real-time adaptability, miniaturization, and signal integrity when deployed on the body. A new Metasurface-Assisted Wearable MIMO Antenna Array is suggested to overcome these difficulties, but it is combined with the deep learning-based control system named Cross Cascaded 3D Contextual Secretary Bird Dilated Convolutional Attention Network (Cross-3C4DA-Net). The most essential innovation is a Cascaded 3D Dilated Convolutional Network (CD-Net) with a Cross-Contextual Attention Mechanism (CCAM), which allows effective spatial and contextual features to be extracted out of dynamic body-centric wireless signals. Cross-3C4DA-Net parameters are also optimized with the Secretary Bird Optimization Algorithm (SBOA) that guarantees convergence faster and better learning accuracy with complex and noisy conditions, such as biomedical conditions. Benchmark WBAN telemetry datasets and simulated antenna channel condition are used to evaluate Cross-3C4DA-Net. The return loss of the proposed antenna array is − 35.72 dB, which means that there is not much reflection and the impedance of the two is well matched. The gain is 8.94 dBi which guarantees high radiation efficiency and direction transmission and Voltage Standing Wave Ratio (VSWR) is kept at a nominal 1.07 which is with acceptable operational range. There are two significant benefits that this strategy provides: improved signal quality and less overhead of the system. It shows that it can be deployed in real-time in wearable Biomedical Telemetry (BT) with minimum energy usage and high reliability.
Resource Allocation (RA) in Device-to-Device (D2D) communication refers to the direct sharing of resources such as spectrum, energy, data, or computational power between nearby devices without routing through a central base station. This paradigm enhances network efficiency, reduces latency, and improves spectrum utilization, making it ideal for applications in 5G and beyond. By enabling devices to collaborate and support each other, resource exchange in D2D networks can offload traffic from the core network and extend connectivity in areas with limited infrastructure. However, it requires intelligent resource allocation, trust management, and interference coordination to ensure reliable and fair communication. This work proposes a new Security Aware RA Framework to sustain D2D links during communication in Long-Term Evolution Advanced systems. Particularly, the network is considered under 2 types of frequency bands namely, Millimeter-Wave as well as microwave tiers. Here, a novel optimal RA model is presented that ensures the security level of users in the network based on missed detections and false alarms. Further, RA is carried out using a new algorithm named as Secretary bird Integrated Hippopotamus optimization algorithm (SBI-HOA) that merges the concepts of Hippopotamus optimization algorithm and Secretary bird optimization algorithm. The optimal RA in D2D facilitates the recurrent update of interfering level and aids in maintaining the link between D2D devices. The optimal RA is performed by taking account of interference, security and sum rate. From analysis, a less risk of 0.1 and a high sum rate of 80,000 s*Hz was obtained by proposed SBI-HOA-based optimal RA framework.
Pressure sensors are essential to ensure the efficiency of the health monitoring system such as rehabilitation and ergonomic application. Compared to traditional health monitoring, miniature size for low power consumption, flexibility, biocompatibility, rapidity, and low-cost sensors are among the essential parameters. Micro-Electro-Mechanical System (MEMS) pressure sensors offer very small size where it can respond rapidly to a very small changes in pressure with high sensitivity and excellent linearity. Selection of piezoresistive material is crucial to enhance the performance of MEMS pressure sensor. Silicon-based MEMS piezoresistive pressure sensor suffers from brittleness and poor scalability. As such, carbon nanomaterial such carbon nanotube and graphene offer promising solutions as piezoresistive materials. To the best of our knowledge, there has been limited findings on the analysis of different type of carbon-based piezoresistive pressure (Graphene and CNT) sensors in terms of change of resistance, linearity and sensitivity through finite element simulation. Hence, the aim of this paper is to compare and analyze two different carbon nano materials (graphene and carbon nanotube) that act as piezoresistive materials for MEMS pressure sensor with combinational Dash E-shape design. Finite element simulation results show that graphene-based MEMS piezoresistive pressure sensor shows high sensitivity of 0.6458 mV/V∙mmHg. As on linearly increasing the output voltage, it is noticed that graphene performs well with 99
A Fe2O3/La2O3 nanocomposite was synthesized by mechanically blending of iron oxide and lanthanum oxide powders, each prepared individually via the co-precipitation method. The resulting binary oxide was employed to fabricate a thick film sensor using a conventional screen-printing technique. Structural characterization by X-ray diffraction (XRD) confirmed the high crystallinity of the nanocomposite with an average crystallite size of 48 nm. Scanning electron microscopy (SEM) revealed a rod-like morphology with an interconnected network, and an average particle size of 158.84 nm. Elemental analysis using EDAX confirmed the presence of Fe, La and O without any detectable impurities. Optical characterization using UV–Visible spectroscopy estimated the optical band gap to be 1.18 eV. FTIR analysis indicated the presence of characteristic Fe–O and La–O bonds, while the Raman spectrum revealed structural features associated with Fe2O3 and La-induced lattice distortions. Photoluminescence (PL) studies showed strong emission peaks at 511 nm and 603 nm, suggesting efficient radiative recombination. The gas sensing performance of the Fe2O3/La2O3 thick film sensor was evaluated towards various gases such as H2S, NO2, ethanol, LPG, CO2 and demonstrated the highest gas response and selectivity towards NO2 gas. A maximum response of 87.6 was observed at 100 ppm NO2 concentration and an operating temperature of 60 °C, with a rapid response and recovery time of 8 s and 9 s respectively. Additionally, the photocatalytic activity of the nanocomposite was tested for the degradation of methyl green (MG) dye under visible light. Optimal degradation efficiency reached 95.45
Early-stage detection of brain tumors remains a significant clinical challenge, requiring sensing platforms capable of resolving subtle variations in the optical properties of biological tissues and fluids. In this work, a multimodal terahertz (THz) surface plasmon resonance (SPR) metasurface biosensor is proposed, integrating a multi-resonator metallic architecture with a vertically stacked graphene/WS₂ two-dimensional heterostructure. This design enables enhanced electromagnetic field confinement and the generation of multiple sharp resonance features within a compact footprint. The sensor is numerically investigated using full-wave COMSOL Multiphysics simulations, demonstrating high sensitivity (up to 1538 GHz/RIU), a quality factor of 44, and a figure of merit of 134, indicating strong spectral selectivity and detection precision. To improve computational efficiency, a Random Forest regression model is developed to predict refractive index variations with high accuracy (≈ 98
The increasing demand for power-efficient and high-performance communication systems has driven advancements in radio frequency micro electro mechanical systems (RF MEMS) technology, particularly in bistable lateral switches. These switches offer superior RF characteristics, including minimal insertion loss with excellent isolation and low power operation, making them ideal for redundancy selection where reliable RF signal switching and uninterrupted performance are critical. The mechanical characterization of such a bistable MEMS switch was conducted at IISc, Bangalore. To make the switch operate at RF frequencies and achieve better RF performance, the sidewall coating of coplanar waveguides (CPW) has been done and the simulation has been carried out in high frequency structure simulator (HFSS). To optimize the CPW sidewall coating length ( L_sm ), an artificial neural network (ANN) model was developed. Variations in the sidewall coating length ( L_sm ), influences the current distribution in the sidewalls and in turn the inductance (L) of the CPW gets modified. In this paper, the RLC (resistor, capacitor, inductor) equivalent circuit model is formulated for the bistable lateral switch. The inductance (L) value obtained for the corresponding optimized CPW sidewall coating length ( L_sm ) from the ANN model is used in the RLC model. The influence of inductance (L) on the RF Performance (S11, S21) is observed using advanced design system EM simulator. The simulation results show an isolation below − 70 dB and insertion loss of − 0.47 dB at 6 GHz for both STABLE STATEs. The S parameters obtained using RLC model agreed well with results obtained from HFSS simulations.
In the recent past, ferrite-based materials received attention in different fields such as energy storage, wastewater treatments, microwaves, electronics, sensors, and magnetic devices. These materials have different advanced physical and chemical features. The ferrite-based electrode materials exhibit good lifespans and high-power densities. The present work demonstrated the electrochemical performances of Mn blended NiFe2O4. The ultrasonic and solid-state methods were utilized in the synthesis of NiFe2O4/Mn-200. The developed NiFe2O4/Mn-200-based electrode material was studied using various analytical methods and could potentially be used as a supercapacitor electrode. The maximum specific capacitance of NiFe2O4 was found to be 120 F/g at the current density of 2 A/g and increased to 450 F/g after integration of Mn. The maximum energy and power densities of NiFe2O4/Mn-200 were evaluated to be 6.7 Wh/kg and 4998 W/kg, respectively.
Wireless Sensor Microsystem Networks represent the infrastructure of next generation Internet of Things, where communication, energy module and computation are incorporated into compact microsystem based sensor nodes. The microsystem nodes enable miniaturized hardware integration, scalable deployment and low power consumption in large scale environments. Meanwhile, due to limited computational resources, efficient clustering and routing mechanism is essential to ensure reliable communication with enhanced network lifetime. This paper proposes a unified hybrid framework to achieve energy aware clustering and intelligent routing in Internet of Things based Wireless Sensor Microsystem Networks. In order to enhance network lifetime and estimate better intra and inter-cluster communication, the Cluster Head selection and clustering strategy are optimized. Initially, the sensor network is clustered using a quantum-inspired fuzzy C-means model that introduces probabilistic perturbation to enhance exploration during cluster formation. Moreover, the crossover-boosted revolutionary optimization algorithm selects the optimal cluster head that integrates exploration–exploitation trade-offs for detecting the energy-efficient cluster nodes. The routing path of the cluster is analyzed by a Dual-Encoder Transformer with an Attention-Guided Energy Awareness framework to predict optimal multi-hop paths thereby minimizing delay and energy consumption during network transmission. The effectiveness of the proposed method is computed using various measures. Experimental results demonstrate that the proposed model achieves a network lifetime of 32,300 rounds, and energy consumption of 35 Joules. Furthermore, the proposed framework outperforms existing methods and other baseline approaches, showing significant improvements in energy efficiency, network lifetime, and data transmission performance.
This work focuses on enhancing mixing in passive U-turn micromixers. Microchannels with three different U-turn geometries namely curved, rectangular, and triangular shapes were examined. The effects of the number of mixing units and the design of U-turn geometries on mixing behaviour were analysed for Reynolds numbers between 0.1 and 60. The microchannels with different U‑turn shapes and fixed axial lengths of 6.8 mm, 8.4 mm, and 10.0 mm were investigated. Mass concentration plots on the x–y and y–z planes at various axial positions were analyzed to provide detailed insights into the mixing behavior and underlying mechanisms. At low Reynolds numbers (Re < 5), mixing is dominated by diffusion; above Re = 5, secondary flows and chaotic advection become the main mechanisms. The microchannel with rectangular-shaped U-turn showed better performance for Re > 5.
The scanning dynamics and the motion nonlinearity of gimbaled and gimbal-less MEMS micromirrors are studied using Euler’s equations of motion for the application to omnidirectional scanning of light detection and ranging. The motion interaction between the rotations around the two axes is dependent intricately on moments of inertia, rotation angles, and angular velocities. The magnitudes of the interaction terms of the equations increase with the increase of the difference of the moments of inertia along the orthogonal axes of mirrors. Solving the motion equations by Runge–Kutta method, we investigate comparatively the omnidirectional trajectories of a gimbaled micromirror and a gimbal-less micromirror with the same moment of inertia. The models of the mirrors are based on the fabricated MEMS micromirror having the resonant frequency of 5336 kHz. The circular trajectory is obtained in a wide range of actuator torque in vacuum at quality factor of 1000 for the gimbaled mirror, while for the gimbal-less mirror, the circular trajectory tends to deform elliptically. The trajectory is stable but occasionally become uncontrollable by the actuator torque. Decreasing the quality factor to 100 corresponding to atmospheric pressure, the elliptical deformations are suppressed although the scan angles decreased. The minimization of the motion nonlinearity is discussed from the viewpoints of damping and actuator torque.
This study investigates the impact of alternatively incorporated ZnO, Si and ZnCdTe layers in the active region of the device, a finding that has not been previously reported in detail. The performance of the Quantum-well based hetero-structure ZnO/Si/ZnCdTe PIN photo-detector has been explained and compared to its GaN/Si/InGaAs counterpart in UV-Visible-NIR wavelength region. The Nonlinear Quantum Modified Drift-Diffusion (QM-D-D) model is employed for electro-optical characteristic studies in the devices, which find application in UV-Visible-NIR detection. The results indicate that the designed hetero-structure ZnO/Si/ZnCdTe PIN photo-detector offers higher external quantum efficiency and photo-responsivity compared to its GaN/Si/InGaAs counterpart in UV-Visible-NIR wavelength region. The noise reduction in the designed photo-detector makes it more suitable for use as low-noise photon detectors. Furthermore, the study evaluates the suitability of 3 × 5 array-based photo-detectors in terms of photo responsivity and external/internal quantum efficiency. The validity of the indigenously developed QM-D-D model is confirmed through experimental verification. In addition, failure analysis and fabrication feasibility of the new class of designed PIN photo-detector is presented in this paper. To the best of the authors’ knowledge, this is the first report on quantum-well based hetero-structure ZnO/Si/ZnCdTe PIN photo-detector which can be used to detect the photon in UV through Visible to NIR range of EM spectrum.
To address the limitations of traditional Gabor filters—specifically their reliance on manual parameter adjustment and limited universality, an iris recognition algorithm based on adaptive Gabor filters and support vector machine (SVM) is proposed. Due to the unique grayscale distribution characteristics of the iris, the combination of adaptive thresholding and least squares methods is used for iris inner edge localization. Based on the characteristics of the center and outer edge of the pupil, an extended ray method is adopted for iris outer edge localization, followed by improvements in the sub-calculus operator to narrow down the search area, thus enhancing the efficiency of iris localization. An improved Particle Swarm Optimization (PSO) algorithm is employed to optimize the selection of Gabor filter parameters, enabling adaptive parameter selection for Gabor filters and completing feature encoding of iris images. By combining the nonlinear search capabilities of PSO and SVM, iris feature matching and recognition are effectively achieved. The improved algorithm demonstrated an increase in recognition rate by 0.59
Pinching-Antenna Systems (PASS) have emerged as a promising low-power alternative to traditional high-power 6G Multiple-Input Multiple-Output (MIMO) architectures by facilitating beamforming through selective antenna activation along dielectric waveguides instead of power-hungry phase shifters. Although PASS can potentially reduce the hardware power consumption of the antenna by over 95
