Chakdaha College

Heart rate variability (HRV) is a physiological indicator that effectively identifies stress and evaluates cardiac health. This study presents a deep learning architecture for resilient HRV signal processing that integrates Convolutional Neural Networks (CNN) and Long Short-Term Memory (LSTM) units enhanced by a Split-Window Approach (SWA). The SWA improves the model's ability to detect transient patterns associated with cardiac irregularities and stress by effectively segmenting HRV data into temporally consistent slices. The proposed CNN-LSTM-SWA model significantly outperforms conventional HRV analysis techniques, as demonstrated by studies on standard HRV datasets. It attains exceptional precision in both stress identification and cardiac condition classification. These findings indicate that real-time applications in wearable and mobile health monitoring systems could be feasible.

Alzheimer's disease (AD) is a progressive neurodegenerative disorder, and a great deal of pathological changes can be observed in the early stages of the disease before significant cognitive function is clinically evident. In this study, a multimodal Transformer framework named NeuroFusion for early and explainable detection of AD is suggested, which integrates structural magnetic resonance imaging (MRI), positron emission tomography (PET), and clinical information. The framework consists of modality-specific encoders to obtain complementary representations, and cross-modal attention to uncover relationships among anatomical, metabolic, cognitive and demographic features. The architecture proposed aims for the classification of cognitively normal individuals, mild cognitive impairment (MCI), AD patients, and for an easy-to-understand patient-level prediction. An explainable artificial intelligence component is used to determine the influential brain regions and clinical parameters using imaging attribution and feature analysis by SHAP. The illustrative evaluation shows that the unimodal and conventional multimodal configurations are outperformed by NeuroFusion, with a ROC-AUC of 0.961, and ablation analysis reveals the contribution of MRI, PET, clinical information and of cross-modal attention. The framework also includes calibration and explainability assessment which enhance the reliability of the framework. In summary, NeuroFusion offers a holistic solution that integrates multimodal learning, early-stage detection, and interpretability, and has the potential to yield more transparent AI-aided Alzheimer's assessment. The numerical results need to be validated through an empirical study on an actual trained model and an independent set of data.

Stress is a conglomerate of changes that can be emotional or physiological in response to disturbances. This paper presents a model based on deep learning, termed ‘CognitiveGAN,’ capable of augmenting the dataset by a factor of 24 and autonomously classifying heart rate variability (HRV) derived from electrocardiogram (ECG) signals into stressed and prestressed conditions. We train the model end-to-end on signal acquisition from 98 subjects under prestressed and stressed conditions. The model augments the preprocessed dataset by 24 times resulting in 11280 training samples, which the classifier model uses to autodiagnose the stressed and normal signals. The suggested network utilizes a separable convolution block-based skip connection framework, drawing inspiration from the ResNet architecture. The 1D HRV signals are converted into 2D spectrograms to utilize the benefits of a 2D convolutional neural network (CNN) along with a generative adversarial network (GAN). The architecture model was found to have an accuracy of 98.18% and an AUC of 0.98, indicating that the cardiac dynamics were healthier under prestressed conditions, thus suggesting that stress management should be ensured. Our model can be used in the embedded systems in various machines in medical practice, which involves the analysis of cognitive stress or its impact on HRVs.

This study presents a detailed investigation of bin-bin correlations in p-Pb collisions at a center-of-mass energy of √(s_NN) = 5.02 TeV using factorial correlators. The analysis examines the dependence of these correlations on pseudorapidity and event centrality, considering most-central to mid-central collision classes ( 0-10%, 10-20%, 20-30%, 30-40%, and 40-50% ), as well as minimum-bias events. The simulated data sets are generated using the Monte Carlo-based A Multi-Phase Transport (AMPT) model (v2.26), employing configurations both with and without the string-melting mechanism. Factorial correlators are evaluated within a pseudorapidity interval of Δη = 12, which is subdivided into 40 and 20 bins corresponding to bin widths of δη = 0.3 and 0.6, respectively. The results reveal strong bin-bin correlations that increase with decreasing bin separation and exhibit no significant dependence on bin width or event centrality. Nevertheless, the observed differences are statistically robust and significant, and the overall behavior, for both string-melting and without string-melting scenarios, deviates markedly from expectations based on the log-normal approximation.

The report focuses on broadband microstrip antennas for telephony applications. It highlights how microstrip antennas, through co-linear patch design, address the challenge of large antenna size by enabling compact designs suitable for GSM technology. The project also demonstrates antenna designs that can operate in both single-band (0.58 GHz) and multiband (0.35 and 0.28 GHz) modes within the same size, and also modify structure of patch to enhance band width 1 1% supporting device miniaturization and efficient wireless communication with 1 6 d B gain.