Volume 14 | Issue 4 |

 Abstract

Diabetes Mellitus (DM) is a chronic metabolic disorder characterized by persistent hyperglycaemia resulting from defects in insulin secretion, insulin action, or both. The global burden of diabetes is rapidly increasing, with approximately 589 million adults affected worldwide in 2024, projected to rise to 853 million by 2050. Among its complications, diabetic foot problems are a major cause of morbidity, often resulting from neuropathy, peripheral arterial disease, and poor wound healing. These complications can lead to ulcers, infections, and even amputations if not properly managed. Adequate knowledge and practice of foot care are essential in preventing such complications, yet factors like illiteracy may hinder effective self-care. The present study aimed to assess the knowledge and practice regarding diabetic foot care among newly diagnosed diabetic patients. A quantitative approach with a cross-sectional descriptive survey design was adopted. The study was conducted in a selected hospital with a sample of 100 patients aged 20-90 years, selected using non-probability purposive sampling. Data were collected using a structured questionnaire and a practice checklist. The findings revealed that the majority of patients (58%) had average knowledge, 35% had good knowledge, and 7% had poor knowledge regarding diabetic foot care. Regarding practice, 64% demonstrated average practice, 27% poor practice, and only 9% good practice. A strong positive correlation (r = 0.929) was observed between knowledge and practice. Educational qualification showed a significant association with both knowledge and practice levels, while most other demographic variables were not significantly associated. The study concludes that although most patients possess moderate knowledge and practice levels, there remains a need for improved education and awareness programs. Targeted interventions, especially for less educated individuals, are essential to enhance diabetic foot care practices and prevent complications.

Licence: creative commons attribution 4.0

  License

Creative Commons Attribution 4.0 and The Open Definition

 Keywords

Diabetes Mellitus, Diabetic Foot, Foot Care, Knowledge, Practice, Neuropathy, Peripheral Arterial Disease, Foot Ulcer, Patient Education, Chronic Disease Management

  License

Creative Commons Attribution 4.0 and The Open Definition

 Abstract

Deepfake technology poses significant threats to digital media authenticity, necessitating advanced detection mechanisms. This paper proposes a hybrid convolutional neural network (CNN) and ResNet architecture that leverages residual learning to enhance feature extraction from facial videos. The framework integrates multi-scale convolutions for low-level texture analysis with deep residual blocks for high-level semantic understanding. Key innovations include a custom fusion layer combining CNN outputs with ResNet skip connections, trained via binary cross-entropy loss on synthetic deepfake datasets. Experiments on benchmark datasets demonstrate superior performance, achieving 98.2% accuracy, 97.5% precision, 98.0% recall, and 97.8% F1-score, outperforming baselines like MesoNet and Xception by up to 12%. Ablation studies confirm the efficacy of residual fusion in mitigating gradient vanishing. This approach advances real-time deepfake mitigation for applications in cybersecurity and media forensics.

Licence: creative commons attribution 4.0

  License

Creative Commons Attribution 4.0 and The Open Definition

 Keywords

Deepfake Detection, CNN, ResNet, Video Forgery Detection Deep Learning, Residual Networks, Cybersecurity

  License

Creative Commons Attribution 4.0 and The Open Definition

 Abstract

Abstract-- Prediction of stock prices has been an important area of research for a long time. While supporters of the efficient market hypothesis believe that it is impossible to predict stock prices accurately, there are formal propositions demonstrating that accurate modeling and designing of appropriate variables may lead to models using which stock prices and stock price movement patterns can be very accurately predicted. Researchers have also worked on technical analysis of stocks with a goal of identifying patterns in the stock price movements using advanced data mining techniques. In this work, we propose an approach of hybrid modeling for stock price prediction building different machine learning and deep learning- based models. For the purpose of our study, we have used NIFTY 50 index values of the National Stock Exchange (NSE) of India, during the period December 29, 2014 till July 31, 2025. We have built eight regression models using the training data that consisted of NIFTY 50 index records during December 29, 2014 till December 28, 2023. Using these regression models, we predicted the open values of NIFTY 50 for the period December 31, 2018 till July 31, 2020. We, then, augment the predictive power of our forecasting framework by building four deep learning- based regression models using long-and shortterm memory (LSTM) networks with a novel approach of walk- forward validation. Using the grid-searching technique, the hyperparameters of the LSTM models are optimized so that it is ensured that validation losses stabilize with the increasing number of epochs, and the convergence of the validation accuracy is achieved. We exploit the power of LSTM regression models in forecasting the future NIFTY 50 open values using four different models that differ in their architecture and in the structure of their input data. Extensive results are presented on various metrics for all the regression models. The results clearly indicate that the LSTM-based univariate model that uses one-week prior data as input for predicting the next week's open value of the NIFTY 50 time series is the most accurate model.

Licence: creative commons attribution 4.0

  License

Creative Commons Attribution 4.0 and The Open Definition

 Keywords

Keywords-- Stock Price Prediction, Regression, Long and Short-Term Memory Network, Walk-Forward Validation, Multivariate Time Series.

  License

Creative Commons Attribution 4.0 and The Open Definition

 Abstract

Licence: creative commons attribution 4.0

  License

Creative Commons Attribution 4.0 and The Open Definition

 Keywords

Key words: Diabetes mellitus, Self-care, Mobile based education

  License

Creative Commons Attribution 4.0 and The Open Definition

 Abstract

Diabetic Retinopathy (DR) is one of the leading causes of preventable blindness worldwide, making early and scalable screening systems critically important. Although deep convolutional neural networks have demonstrated strong perfor mance in retinal image classification, fully training or fine-tuning such models is often computationally demanding and difficult to deploy in resource-constrained environments. In this work, we propose a hybrid classification framework that combines deep feature extraction with classical machine learning to achieve a balance between predictive performance and computational efficiency. A pretrained ResNet-50 network is used as a fixed feature extractor to obtain 100,352-dimensional representations from 5,000 retinal fundus images from the APTOS 2019 dataset. To reduce redundancy and computational overhead, Principal Component Analysis (PCA) with randomized Singular Value Decomposition is applied to compress the feature space to 1,024 principal components, retaining approximately 95-97% of the cumulative explained variance. The reduced features are standardized and evaluated using multiple classifiers, including Logistic Regression, Random For est, XGBoost, LightGBM, CatBoost, and a stacking ensemble. Among these, the stacking ensemble achieved the best perfor mance, reaching an accuracy of 91.5%, a weighted F1-score of 0.9127, and a macro ROC-AUC of 0.9675. Compared with a f ine-tuned EfficientNet baseline, the proposed framework signifi cantly reduces model complexity and training requirements while maintaining competitive classification performance. Additionally, SHAP-based analysis is employed to enhance interpretability and provide insight into model decision behavior. Overall, the proposed hybrid framework demonstrates that computationally efficient deep-feature compression combined with ensemble learning can serve as a practical and interpretable solution for automated DR screening.

Licence: creative commons attribution 4.0

  License

Creative Commons Attribution 4.0 and The Open Definition

 Keywords

Diabetic Retinopathy, Retinal Fundus Images, Deep Feature Extraction, Transfer Learning, ResNet-50, Prin cipal Component Analysis (PCA), Machine Learning Classifica tion, Hybrid Modeling, Class Imbalance Handling, Explainable Artificial Intelligence (XAI), SHAP

  License

Creative Commons Attribution 4.0 and The Open Definition

 Abstract

This paper examines the transformation and continuity of Bengal's middle-class intelligentsia in post-independence Kolkata, tracing its intellectual lineage to the Bengal Renaissance. Under colonial rule, the bhadralok emerged as a culturally dominant and intellectually influential class, shaping discourses in social reform, literature, science, and politics. However, the Partition of 1947 constituted a decisive rupture, producing demographic upheaval, economic dislocation, and a reconfiguration of Kolkata's urban and social fabric. The influx of refugees, combined with industrial decline, weakened the material base and cultural hegemony of this class. In the decades following independence, the intelligentsia underwent a complex process of continuity and crisis. While they maintained a significant presence in cultural production--particularly in literature, theatre, cinema, and political thought--their earlier role as national leaders diminished within an increasingly centralized postcolonial state. This marginalization, alongside rising unemployment and economic stagnation, fostered disillusionment with state-led development. Consequently, sections of the intelligentsia gravitated toward Marxist and leftist ideologies, culminating in their engagement with radical movements such as the Naxalite uprising. This phase revealed both ideological commitment and internal contradictions, as revolutionary aspirations often gave way to critical reassessment. The paper further explores the impact of economic stagnation, professional insecurity, and intellectual out-migration, which fragmented the coherence of the bhadralok identity. The liberalization era introduced new middle-class aspirations shaped by consumerism and globalized values, challenging earlier intellectual traditions. Despite these shifts, elements of critical inquiry, secularism, and cultural engagement persist. The study argues that Bengal's middle-class intelligentsia represents a dynamic interplay of legacy and adaptation, offering a crucial lens for understanding postcolonial urban and intellectual transformations in India.

Licence: creative commons attribution 4.0

  License

Creative Commons Attribution 4.0 and The Open Definition

 Keywords

Bengal Renaissance,Partition of Bengal,Middle-Class Intelligentsia, Post-Independence Dislocation,postcolonial transformation.

  License

Creative Commons Attribution 4.0 and The Open Definition

 Abstract

In rural schools in India, the process of tracking the attendance of students is still mostly manual, with teachers calling out students or handing out paper registers. It is a slow, unreliable and manipulated method of assessment- a student can have a friend mark him/her present even when he/she is not even in school. In the long run, the wrong records cause issues when examining students and also complicate the work of the administrators to identify students who are skipping classes. The following paper outlines an attendance management system which tracks the attendance of students automatically with the help of a webcam and a face recognition which does not require any costly hardware or an internet connection that is fast enough to support cloud-based AI processing. The recognition engine is completely implemented within a web-based application based on the HTML5 Canvas API - 25 quick webcam snapshots are taken during registration, a compact 128-number representation of the face of the student is created, and any new snapshots are compared with existing representations by a scoring formula based on two complementary distance measures. The testing on sixty students in three lighting conditions revealed that the system was accurate in identifying the student's 96.5 percent of the time in normal classroom lighting and 84.7 percent of the time during harsh window light illuminating students, and that each recognition check took approximately 321 milliseconds. The entire site, web interface, server, and reporting tools, has been developed in TypeScript, React, Express, and PostgreSQL and can be accessed on a cloud server at 10 Mbps, which is the maximum bandwidth most rural schools in India have under the BharatNet programme. Teachers are also able to manually mark attendance where necessary, download records in spreadsheet files as well as simple charts that indicate classes with low attendance. The paper captures the technical details, test results and deployment in four tables. Keywords: face recognition, school attendance, rural India, webcam, HTML5 Canvas, hybrid similarity, React, Express, PostgreSQL, TypeScript, BharatNet

Licence: creative commons attribution 4.0

  License

Creative Commons Attribution 4.0 and The Open Definition

 Keywords

face recognition, school attendance, rural India, webcam, HTML5 Canvas, hybrid similarity, React, Express, PostgreSQL, TypeScript, BharatNet, Drizzle ORM, Zod validation, cosine similarity, Euclidean distance, 128-number face descriptor, browser-based encoding, 96.5% accuracy, 10 Mbps deployment, Digital Personal Data Protection Act 2023, Ajeenkya DY Patil University, proxy attendance, manual fallback, dashboard charts, rural schools

  License

Creative Commons Attribution 4.0 and The Open Definition

 Abstract

Green Consumerism is acquisition of a product in a market in sustainable way and eco- friendly environment. Now a days, our society is very much worried about our environment and the pollution caused by us. So, in order to protect our environment, we should follow green practice. Due to pollution and decomposition of Industrial waste, the ozone layer is getting depleted. The organization has to step forward to protect the environment by producing and selling. Moreover, because of this green product the sales of the organization can also be increased and it protects our environment. The questionnaire was prepared and sent to 100 people but, in that 50 respondents given their opinions. According to the study, the Vouch Factor, Environmental Factor and Reference Group are the factors influencing the consumer to Buying Behaviour Green Products

Licence: creative commons attribution 4.0

  License

Creative Commons Attribution 4.0 and The Open Definition

 Keywords

Green Packing, Vouch Factor, Buying Behaviour, Green Consumerism.

  License

Creative Commons Attribution 4.0 and The Open Definition

 Abstract

Traffic congestion at busy intersections is a major issue in urban transportation systems, leading to increased travel time, fuel consumption, and environmental pollution. This study focuses on optimizing traffic flow at highly congested intersections using advanced traffic management techniques. The research analyzes traffic patterns, vehicle density, and signal timing through data collection and simulation methods. Various strategies such as adaptive traffic signal control, lane management, and intelligent transportation systems (ITS) are evaluated. The proposed model aims to reduce vehicle delay, minimize queue length, and improve overall traffic efficiency. The results demonstrate that optimized signal timing and real-time traffic monitoring significantly enhance traffic flow and reduce congestion levels. This study provides practical insights for urban planners and engineers to develop efficient and sustainable traffic management solutions.

Licence: creative commons attribution 4.0

  License

Creative Commons Attribution 4.0 and The Open Definition

 Keywords

Traffic Flow Optimization, Traffic Congestion, Signal Timing, Adaptive Traffic Control, Intell Transportation System

  License

Creative Commons Attribution 4.0 and The Open Definition

 Abstract

This is because correct predictions of residential electricity usage are essential in reducing costs and management of energy sustainably. The conventional billing systems offer limited information on the future consumption trends and do not offer much proactive optimization. In this paper, I will introduce a Smart Energy Auditor and Electrical Bill Prediction System using an interferon-free, deep learning, based system to make predictions on monthly electricity bills and allow energy planning within a budget constraint. The offered system employs a neural network regression model with Keras to use some appliance-level consumption data obtained by mathematical disaggregation, and therefore, it does not require intrusive hardware sensors. The design calculates energy use based on standard electrical power equations and uses a new budget-to-time optimization unit that carries out an inverse calculation to find out the allowable hours of appliance operation within any financial limit. Exposure to experimentation proves that the system yields high prediction accuracy with a mean absolute error of 3.2 per cent and root mean squared error of 4.8 per cent, and is effective in reducing the cost of energy through intelligent usage recommendations. This is due to the fact that the non-invasive nature, the real-time predictive, and user-friendly optimization functionalities makes this system an effective and scalable solution to residential energy management.

Licence: creative commons attribution 4.0

  License

Creative Commons Attribution 4.0 and The Open Definition

 Keywords

Electricity Bill Prediction, Energy Management System, Deep Learning, Non-Intrusive Load Monitoring, Energy Consumption Forecasting, Budget Optimization, Keras, Neural Networks, Smart Home Energy

  License

Creative Commons Attribution 4.0 and The Open Definition