Rayalaseema University

Climate change has emerged as a dominant driver of global biodiversity loss, profoundly altering zoological diversity, species distribution, and ecosystem functioning. Rising temperatures, erratic precipitation, ocean acidification, and habitat fragmentation disrupt physiological processes, migration patterns, and trophic interactions across taxa. This review explores the impact of climate change on zoological diversity and species distribution. Terrestrial, aquatic, and marine ecosystems exhibit varying degrees of vulnerability, with endemic and specialised species facing heightened extinction risks. Polar fauna such as seals and polar bears, tropical primates, coral reef communities, and migratory birds demonstrate clear evidence of climate-induced population decline and range shifts. Evolutionary responses, including genetic adaptation and phenotypic plasticity, are observed in some taxa but remain insufficient to counter rapid environmental changes. Advances in remote sensing, GIS-based species modelling, and molecular techniques such as genomics and environmental DNA (eDNA) have enhanced understanding of climate impacts and enabled predictive biodiversity mapping. Conservation measures encompassing ecosystem-based adaptation, habitat restoration, captive breeding, and wildlife corridor establishment contribute to mitigating biodiversity loss. Policy frameworks under the Convention on Biological Diversity (CBD), IUCN, and IPCC emphasise integrating nature-based solutions with national climate strategies. Despite significant progress, uncertainties in predictive models, data scarcity, and limited integration between ecological and socio-economic systems impede effective conservation planning. Future research must prioritise long-term ecological monitoring, multi-scale data synthesis, and cross-sectoral policy implementation to enhance resilience. Integrating indigenous ecological knowledge with scientific innovations will strengthen adaptation measures and ensure sustainable biodiversity management in the face of accelerating climatic shifts.

Brain tumors present a significant diagnostic challenge due to their complex structural variations and subtle visual cues in medical imaging. To address these challenges, this study introduces a novel feature fusion-driven lightweight Convolutional Neural Network (CNN) architecture enhanced with a multi-attention mechanism for accurate and efficient brain tumor detection. The proposed model integrates complementary features extracted from multi-scale convolutional layers and fuses them to capture both global contextual information and fine-grained local details. Furthermore, a multi-attention module is incorporated to adaptively focus on salient regions, improving the model's interpretability and discrimination capability. This approach significantly reduces the computational complexity compared to traditional deep networks, enabling deployment in resource-constrained environments without compromising detection accuracy. Our proposed approach achieved an accuracy of 99.23 %. Comprehensive experiments on benchmark brain tumor MRI dataset determine that our model achieved high performance in terms of accuracy, sensitivity, and specificity, outperforming several state-of-the-art techniques. The promising result suggest that this lightweight, attention-guided feature fusion framework has strong potential for practical, real-time brain tumor screening applications.

In this paper, the design of a small-size and reconfigurable circular patch antenna is introduced for Wireless Sensor Network (WSN) applications with PIN diode switches to achieve frequency agility. The effective current path on the radiating patch is changed by switching ON/OFF the states of some select PIN diodes, and the antenna is switched among several resonant frequencies without modifying any physical dimensions. The antenna design was simulated in ANSYS HFSS, and its performance analyzed in relation to return loss, bandwidth, and radiation pattern. The simulated results for the antenna demonstrate that it has a return loss of below - 10 dB in the selected frequency bands, with consistent radiation characteristics, and can therefore be well suited for low-power adaptive WSN nodes which require reliable multiband communication and minimized hardware complexity.

Cardiovascular disease remains one of the leading causes of death worldwide and there is an imperative need to have strong early-identification strategies. This paper makes a rigorous comparison of the five commonly used machine learning algorithms- Logistic Regression, Decision Tree, Random Forest, Support Vector Machine and K-Nearest Neighbors used to analyze the UCI Heart Disease dataset. The preprocessing of the dataset involved the traditional method of scaling and the separation of the data into training and testing samples at 70:30 proportion. They were then used to determine predictive efficacy in terms of accuracy, precision, recall, and F1-score. Random Forest performed the best in predictive accuracy, recording 90.16 %. Logistic Regression and SVM were also really close to each other in terms of performance with accuracies of 85.25 % and 81.97% respectively. The figures obtained in Decision Tree and KNN were 78.68 % and 75.41 in terms of accuracy respectively. The results, therefore, demonstrate that ensemble-based classifiers, especially Random forest, have better performances in clinical settings, and support the prevailing effect selection of algorithm in healthcare analytics. Due to its shown robustness and ability to support complex data interactions, Random Forest becomes a potentially highly effective model to use in a real-world scenario. Later research could usefully look into hybrid models or deep-learning techniques that would succeed further in terms of diagnostic reliability.

In recent years, privacy-preserving medical image segmentation has become vital for confirming safety and dependability of multi-institutional collaborations. However, traditional models often fail maintain segmentation accuracy owing to noise injection and restricted parameter exchange. Hence, this research proposes a novel framework called secure aggregation and homomorphic entropy-based federated learning (SAHE-FL) for Brain Tumor (BT) segmentation. Magnetic Resonance Imaging (MRI) scans were collected from Brain Tumor Segmentation (BraTS)2020 dataset. The collected data were then preprocessed using intensity normalization, bounding-box cropping, and uniform volume resizing, followed by widespread data augmentation to improve generalization. Subsequently, the processed data were fed into 3Dimensional Fully Convolutional Neural Network (3D U-Net) architecture to extract both local tumor boundaries and global contextual features from various MRI modalities. In addition, SAHE-FL incorporates secure aggregation with homomorphic encryption, which confirms client model updates remain encrypted and masked throughout training, thereby removing need for artificial noise while preserving accuracy. Finally, a secure aggregation mechanism combined with homomorphic encryption safeguards that only aggregated updates was revealed providing reliable privacy protection without reducing model accuracy. Experimental results show that proposed SAHE-FL model attained superior results in terms of sensitivity 0.90% than existing Differential Privacy and Partial Weight Sharing-based Federated Learning (DP-PWS-FL) model.

As CVDs are now a leading cause of death all over the world, it is crucial to focus on efficient and accurate risk prediction models for timely diagnosis and treatment. This work implements a Random Forest Classifier for CVD risk prediction based on a dataset of clinical characteristics – age, BMI, blood pressure, cholesterol – and life style factors – smoking status, alcohol consumption, physical activity. For the purpose of performing stout evaluation, Training set and test set are splits from the dataset, then ensured that all continuous variables are made of same magnitude and all categorical variables are converted to numerations.The computed data is then employed to both build and assess the Random Forest Classifier, which is acclaimed to its ability to handle various datasets and overlook data over fitting. Using confusion matrix, model evaluation results are determined along with accuracy besides offering classification report that gives higher prediction information on high-risk and low-risk categories. Such findings clearly illustrate that, when incorporating the important features and highly correlated variables, the proposed model is ready to distinguish between individuals with various cardiovascular risk levels.In this work, potentials of machine learning in healthcare applications were demonstrated particularly in identification of such patients that need medical attention. The results provide the basis for the future development of the methods in risk factor analysis of cardiovascular or any other disease and underline the significance of the data analysis for solutions of the world’s health challenges.