Volume 14 | Issue 7 |

 Abstract

Oral sedation plays a pivotal role in pediatric dentistry by facilitating the management of dental anxiety, fear, and uncooperative behavior, thereby enabling the successful completion of dental procedures. A variety of pharmacological agents have been employed for this purpose, including benzodiazepines, nonbenzodiazepine GABA agonists, antihistamines, melatonin, ketamine, chloral hydrate, and ?2-adrenergic agonists such as dexmedetomidine. Among these, midazolam remains the most widely utilized oral sedative due to its favorable safety profile, rapid onset, predictable pharmacokinetics, and effective anxiolytic and amnestic properties. Emerging agents such as dexmedetomidine have demonstrated promising results, offering effective sedation with minimal respiratory depression and improved postoperative recovery. This review provides a comprehensive overview of the administration protocols, safety considerations, mechanisms of action, pharmacokinetics, dosage recommendations, clinical efficacy, adverse effects, and reversal agents associated with oral sedatives used in pediatric dentistry. Understanding the advantages and limitations of each agent is essential for optimizing patient safety, enhancing treatment outcomes, and ensuring evidence-based sedation practices in contemporary pediatric dental care.

Licence: creative commons attribution 4.0

  License

Creative Commons Attribution 4.0 and The Open Definition

 Keywords

Pediatric Dentistry; Oral Sedation; Midazolam; Dexmedetomidine; Behavior Management.

  License

Creative Commons Attribution 4.0 and The Open Definition

 Abstract

Indian Knowledge Systems (IKS) signify the rich cultural, ecological and traditional insight developed by indigenous communities over generations. The Irular tribe of Tamil Nadu possesses valuable indigenous knowledge associated to environmental protection, herbal medicine, sustainable living, traditional livelihoods and cultural heritage. But, rapid transformation and changing socio-cultural conditions have challenged the preservation and transmission of this knowledge among younger generations. The present study examined the level of Indian Knowledge Systems and Indigenous Practices among Irular tribal students in Thiruvallur District, Tamil Nadu. A mixed-method research design was adopted and data were collected from 150 Irular tribal students selected through purposive sampling from government and government-aided schools and colleges. The questionnaire developed and validated by Mrs.G.Sangeetha, questionnaire comprising dimensions of Indian Knowledge Systems and Indigenous Practices. Descriptive statistics, independent sample t-tests, analysis were employed for data analysis. The findings revealed that there were no statistically significant differences in Indian Knowledge Systems and Indigenous Practices with respect to gender, age, educational status and occupation of the respondents. The study indicates that awareness and practice of indigenous knowledge are comparatively even among Irular tribal students.

Licence: creative commons attribution 4.0

  License

Creative Commons Attribution 4.0 and The Open Definition

 Keywords

Indian Knowledge Systems, Indigenous Practices, Tribal Students, Cultural Sustainability, Indigenous Knowledge and NEP 2020.

  License

Creative Commons Attribution 4.0 and The Open Definition

 Abstract

Artificial Intelligence (AI) has emerged as a transformative technology across various sectors, particularly in banking and finance. AI-driven applications are increasingly being used for financial planning, investment management, risk assessment, fraud detection, and customer service. The integration of AI in financial decision-making processes enhances efficiency, accuracy, and personalization while reducing operational costs. Despite these advantages, concerns regarding data privacy, algorithmic bias, cybersecurity threats, and ethical issues continue to challenge its widespread adoption. This paper reviews the role of Artificial Intelligence in financial decision-making, examines the major applications of Artificial Intelligence in the financial sector, identifies opportunities and challenges, and discusses future prospects. The study is conceptual in nature and is based on secondary data collected from books, research articles, reports, and scholarly publications. The findings indicate that AI has significant potential to improve financial decision-making. The study concludes that responsible implementation, effective governance, and appropriate regulatory oversight are essential to maximize the benefits of Artificial Intelligence in financial decision-making.

Licence: creative commons attribution 4.0

  License

Creative Commons Attribution 4.0 and The Open Definition

 Keywords

Artificial Intelligence, Financial Decision-Making, Banking, Machine Learning, Investment Management, FinTech

  License

Creative Commons Attribution 4.0 and The Open Definition

 Abstract

Violence against women (VAW) is a global human rights crisis that acts as a structural barrier to sustainable development. This paper employs a qualitative document analysis methodology, synthesizing data from the National Crime Records Bureau (NCRB), the National Family Health Survey (NFHS-5), and academic literature to examine the landscape of gender-based violence in India. The analysis reveals a 15.3% rise in registered crimes against women in 2021, though systemic factors like the "Principal Offence Rule" and the "culture of silence" lead to significant underreporting. Findings highlight a "gender paradox" where high empowerment indicators, such as literacy in Kerala, do not always correlate with reduced violence. The study also identifies emerging threats, including technology-facilitated abuse on dating apps and heightened vulnerability during natural disasters. Ultimately, the research argues that while legal frameworks have modernized, true progress requires a fundamental socio-cultural transformation to dismantle patriarchal norms.

Licence: creative commons attribution 4.0

  License

Creative Commons Attribution 4.0 and The Open Definition

 Keywords

Violence against women, Gender-based violence, Underreporting, Patriarchy and Technology-facilitated abuse

  License

Creative Commons Attribution 4.0 and The Open Definition

 Abstract

The growing dependence of modern power systems on digital communication, smart meters, and automated control has improved the efficiency and reliability of electricity distribution while simultaneously exposing smart grids to abnormal events such as equipment faults, false data injection, electricity theft, and irregular consumption behaviour. Early and accurate detection of such anomalies is essential for maintaining stable, secure, and uninterrupted power supply. This paper presents an optimized Convolutional Long Short-Term Memory (ConvLSTM) model for detecting abnormal events in smart grid systems. The model couples the spatial feature-extraction strength of Convolutional Neural Networks (CNN) with the temporal sequence-learning capability of Long Short-Term Memory (LSTM) networks, enabling it to analyse time-series smart meter readings identified by consumer number (CONS_NO) and labelled as normal or attack (FLAG). The dataset is prepared through a structured pipeline comprising linear interpolation for missing values, quantile-based winsorization for outlier suppression, a three-point moving-average smoothing of consumption sequences, standard scaling, and a custom synthetic oversampling routine to correct class imbalance before an 80:20 stratified train-test split. The ConvLSTM network consists of a one-dimensional convolutional block (128 channels, kernel size 5) with batch normalization and max-pooling, followed by two stacked LSTM layers and a fully connected classification head trained with binary cross-entropy loss and the Adam optimizer. Model hyperparameters are further refined through an automated Optuna-based search across convolutional channels, LSTM hidden size, depth, dropout rates, learning rate, and batch size, with early stopping used to curb overfitting. The trained and optimized system, deployed through an interactive monitoring dashboard, reports an overall test accuracy of 91.35%, with a precision of 92.10% for the normal class, a recall of 91.80% for the attack class, and an F1-score of 91.32%, alongside consistent performance on independent batch evaluations. The results indicate that jointly modelling spatial and temporal dependencies allows the proposed system to reliably distinguish normal consumption from abnormal events, offering a practical pathway toward strengthening the security and resilience of smart grid infrastructure.

Licence: creative commons attribution 4.0

  License

Creative Commons Attribution 4.0 and The Open Definition

 Keywords

Anomaly Detection, Smart Grid, ConvLSTM, Deep Learning, False Data Injection, Energy Theft, Time-Series Analysis, Hyperparameter Optimization, Cyber Security.

  License

Creative Commons Attribution 4.0 and The Open Definition

 Abstract

Smart agriculture has undergone a paradigm shift from manual observation-based farming to data-driven, AI-assisted precision crop management. This literature review systematically examines the evolution of plant growth monitoring systems, encompassing IoT sensor architectures, edge computing frameworks, soil moisture sensing technologies, multivariate sensor fusion, time-series-based growth modeling, satellite-derived vegetation indices, and generative AI integration. Drawing upon fifteen high-quality references spanning 2020-2025, this review identifies critical limitations in existing approaches--including dependence on threshold-based control, inadequate sensor longevity, and absence of unified validation mechanisms--and articulates a clear research gap. The proposed system, designed for M. Tech research, integrates ESP32-based edge intelligence, capacitive soil moisture sensing, asynchronous 5-15 minute data acquisition, Python analytics, Gemini API-based AI interpretation, NDVI satellite validation, and a real-time Next.js dashboard within a hybrid Edge + Cloud architecture. This review positions the proposed framework as a significant advancement toward predictive, scalable, and intelligent plant growth monitoring suitable for Indian agricultural conditions.

Licence: creative commons attribution 4.0

  License

Creative Commons Attribution 4.0 and The Open Definition

 Keywords

Plant Growth Monitoring, IoT, ESP32, Edge Intelligence, Soil Moisture Sensing, Time-Series Analysis, NDVI, Gemini AI, Precision Agriculture, Next.js Dashboard

  License

Creative Commons Attribution 4.0 and The Open Definition

 Abstract

Mental illness prediction from text uses Natural Language Processing (NLP) and deep learning to identify mental health conditions from written content such as social media posts. Traditional classification models, including Logistic Regression and Naive Bayes, often struggle to understand the emotional context and metaphorical expressions commonly found in mental health-related communication, since they rely on simple feature extraction methods such as TF-IDF and assume independence between words. To address this limitation, this paper proposes a hybrid transformer-based architecture that combines MentalBERT, MelBERT, and Convolutional Neural Networks (CNN) for multiclass mental illness prediction. MentalBERT captures domain-specific language patterns associated with mental health, MelBERT interprets metaphorical and figurative expressions, and CNN layers extract deeper spatial features from the resulting embeddings before classification. The model is trained and evaluated on a balanced dataset of 40,000 social media text samples drawn from Reddit communities and a PTSD corpus, covering four classes: Depression, Anxiety, Borderline Personality Disorder (BPD), and Post-Traumatic Stress Disorder (PTSD). Experimental results show that the proposed hybrid model achieves a test accuracy of 92%, precision of 93%, recall of 92%, and an F1-score of 92%, with a best validation accuracy of 93.63% recorded during training, outperforming baseline models such as Logistic Regression, Naive Bayes, and standalone transformer architectures. The system is deployed through a web-based interface that returns the predicted condition together with a confidence score and class probability distribution, supporting its potential use in social media monitoring and clinical text analysis for early detection of mental health risks.

Licence: creative commons attribution 4.0

  License

Creative Commons Attribution 4.0 and The Open Definition

 Keywords

Mental Illness Prediction, MentalBERT, MelBERT, Convolutional Neural Network, Transformer Architecture, Natural Language Processing, Multiclass Classification, Social Media Text Analysis, Deep Learning.

  License

Creative Commons Attribution 4.0 and The Open Definition

 Abstract

Secondary education in India extends beyond the acquisition of academic knowledge; it is a formative stage that shapes character, nurtures personal growth, and enables students to discover their unique strengths and aspirations. Educational aspiration refers to the goals and expectations students set for their future academic and professional lives. Academic achievement, on the other hand, reflects their actual performance in school, measured through grades, test scores, and overall learning outcomes. This study aims to address the level of educational aspiration and academic achievement of secondary school students with respect to gender in Barasat subdivision of North 24pargana in India and find out whether there is any relationship between educational aspiration and academic achievement. The present study was conducted with a sample of 200 students enrolled in West Bengal Board schools. Educational Aspiration Inventory developed by Dr. Yasmin Ghani Khan was administered. Academic achievement was measured using the percentage of marks obtained by the students in the Class IX annual examination.The findings of the study revealed that there is no significant difference of the level of Educational Aspirations between male and female secondary school students but in case of academic achievement the results are inverse. However, a positive and significant relationship exists between academic achievement and educational aspiration.

Licence: creative commons attribution 4.0

  License

Creative Commons Attribution 4.0 and The Open Definition

 Keywords

Academic achievement, educational aspiration, secondary school students, gender

  License

Creative Commons Attribution 4.0 and The Open Definition

 Abstract

Deepfake technology -- built on generative adversarial networks, autoencoders and, increasingly, diffusion-based synthesis -- has moved from a novelty to a genuine threat to digital media integrity, personal privacy and public trust. Between 2018 and 2020 alone, the volume of deepfake video content online grew by roughly 968%, and little suggests the pace has slowed since. This growth has pushed automated detection from a research curiosity into an operational necessity. This paper reports a systematic review of 20 peer-reviewed studies published between 2022 and 2026 and drawn from IEEE Access, IEEE Transactions on Information Forensics and Security, IEEE Transactions on Consumer Electronics, IEEE Transactions on Neural Networks and Learning Systems, Scientific Reports, and MDPI Applied Sciences. Of more than 200 candidate articles identified across five databases, 20 survived a three-round screening process built around explicit inclusion and exclusion criteria. Eighteen of the twenty propose and evaluate a detection method directly; the remaining two are themselves surveys, retained for context. Each study was examined for its architecture, dataset, task framing and reported performance, and the findings were synthesised into six recurring gaps: an almost universal reliance on binary real-versus-fake classification, the near-total absence of per-class performance reporting, limited explainability, weak cross-dataset generalisation, a dependence on video input that rules out single-image forensics, and reliance on datasets that are difficult to reproduce. In response, we outline an EfficientNet-B3 architecture augmented with a Convolutional Block Attention Module (CBAM) and framed as a six-class manipulation-identification task on FaceForensics++ C23 rather than a binary one, and show how each design choice maps back to one of the six gaps identified.

Licence: creative commons attribution 4.0

  License

Creative Commons Attribution 4.0 and The Open Definition

 Keywords

Deepfake detection, EfficientNet, CBAM, FaceForensics++, transfer learning, face manipulation, systematic review, attention mechanism, Grad-CAM, forensic analysis.

  License

Creative Commons Attribution 4.0 and The Open Definition

 Abstract

Loan approval and credit risk assessment are central tasks in the banking sector, where financial institutions must evaluate large volumes of applicant data while minimizing the risk of default. Traditional rule-based and manually evaluated approval processes are time-consuming, inconsistent, and prone to human bias, and they struggle to scale efficiently with the growing number of loan applications and with imbalanced approval/rejection data. This paper presents a hybrid Machine Learning (ML) and Deep Learning (DL) framework for automated loan approval prediction. The proposed system applies classical algorithms, namely Logistic Regression, Decision Tree, Random Forest, Support Vector Machine (SVM), and K-Nearest Neighbors (KNN), alongside deep learning architectures including an Artificial Neural Network (ANN) and residual and densely connected convolutional architectures (ResNet and DenseNet) adapted for tabular financial data. Class imbalance in the dataset is addressed using the Synthetic Minority Over-sampling Technique (SMOTE), and an ensemble Voting Classifier combining Decision Tree and Random Forest predictions is used to improve robustness. The trained system is deployed through an interactive web-based interface that accepts applicant details and returns a real-time loan approval decision together with a risk probability score. On a dataset of 252,000 loan applications, the evaluated models achieved test accuracies exceeding 98% for DenseNet, above 96% for ResNet, above 95% for Random Forest, and above 85% for the Logistic Regression baseline, indicating that deep learning architectures, particularly DenseNet and ResNet, achieved the strongest predictive performance, with ensemble learning also providing a substantial improvement over the conventional linear baseline. The proposed framework reduces manual effort, improves consistency, and supports faster, data-driven decision-making in loan approval processes.

Licence: creative commons attribution 4.0

  License

Creative Commons Attribution 4.0 and The Open Definition

 Keywords

Loan Approval Prediction, Machine Learning, Deep Learning, Credit Risk, SMOTE, Random Forest, ResNet, DenseNet, Voting Classifier, Banking.

  License

Creative Commons Attribution 4.0 and The Open Definition