		FeedBack \| Contact Us \| Links \| Site Map

Welcome to Journal of Theoretical and Applied Information Technology

Submit Paper / Call for Papers

Journal receives papers in continuous flow and we will consider articles from a wide range of Information Technology disciplines encompassing the most basic research to the most innovative technologies. Please submit your papers electronically to our submission system at http://jatit.org/submit_paper.php in an MSWord, Pdf or compatible format so that they may be evaluated for publication in the upcoming issue. This journal uses a blinded review process; please remember to include all your personal identifiable information in the manuscript before submitting it for review, we will edit the necessary information at our side. Submissions to JATIT should be full research / review papers (properly indicated below main title).

Journal of Theoretical and Applied Information Technology
August 2025 | Vol. 103 No.15

Title:	TRAFFIC SPEED PREDICTION AND CONGESTION LEVEL IDENTIFICATION USING FLOATING CAR DATA
Author:	FATEMEH AHANIN, NORWATI MUSTAPHA, MASLINA ZOLKEPLI, NOR AZURA HUSIN
Abstract:	Traffic congestion poses significant challenges to urban mobility and transportation infrastructure worldwide. Accurate prediction of traffic speed and timely identification of congestion levels are crucial for effective traffic management and planning. Owing to the widespread adoption of telecommunication technologies, various traffic datasets have become available, such as Floating Car Data (FCD), which collect real-time information from vehicles in transit, providing a rich and dynamic dataset for analyzing traffic speed. However, predicting traffic speed and identifying congestion levels using FCD remains challenging due to the complexities of traffic dynamics and the non-linear nature of traffic flow. In response, multiple solutions have been proposed using deep learning methods. This study addresses the persistent issue of FCD data sparsity and its limitations in providing consistent, accurate traffic speed predictions. The present work focuses on constructing an LSTM-based method, called LSTM-C, to predict traffic speed. In the proposed LSTM-C method, a new Contrast measure is introduced and incorporated to enhance the prediction of traffic speed across candidate road segments. The LSTM-C model demonstrates a significant improvement in both prediction accuracy and congestion level identification, outperforming existing models such as those by Majumdar et al. and Gao et al. Subsequently, traffic rules are applied to the predicted speeds to determine congestion levels for each segment. The experimental results demonstrate that the proposed model achieves a high level of accuracy, reaching up to 96.697%, which represents an improvement of 1.6% and 1.79% in accuracy compared to the two benchmark LSTM methods employed for speed prediction.
Keywords:	Traffic Speed Prediction, Short-Term Speed Prediction, The Long Short-Term Memory (LSTM), Deep Learning, Data-Driven Traffic Analysis
Source:	Journal of Theoretical and Applied Information Technology 15^th August 2025 -- Vol. 103. No. 15-- 2025
Full Text

Title:	VRBR-NET: A NOVEL DEEP LEARNING MODEL FOR THERMOGRAPHIC BREAST CANCER DETECTION
Author:	ANUSHA DERANGULA, TAN KUAN TAK, R. THIAGARAJAN, PRAVIN RAMDAS KSHIRSAGAR
Abstract:	Breast cancer remains a leading cause of mortality worldwide, and early, accurate detection is critical for effective treatment and improved patient outcomes. This paper addresses the challenge of advanced breast cancer detection by introducing IVRBR-Net, a novel deep learning model that integrates Inception V4, ResNet-50, and Bidirectional Recurrent Neural Networks (RNNs) to analyze thermographic images. The methodology begins with rigorous preprocessing steps—grayscale conversion, contrast enhancement via Multipurpose Beta Optimized Bi-histogram Equalization (MBOBHE), noise reduction through bilateral filtering, and image refinement using the Affine Projection Algorithm (APA)—to optimize image quality while preserving essential features. Feature extraction is performed using Scale-Invariant Feature Transform (SIFT) and Haralick descriptors, capturing critical statistical properties such as mean, variance, entropy, and skewness. The model is trained and validated on a dataset of 120 high-resolution thermographic images from the Database for Mastology Research (DMR). IVRBR-Net achieves outstanding performance metrics, including an accuracy of 99.82%, specificity of 99.74%, sensitivity of 99.35%, precision of 99.48%, and an F1-score of 99.68%, significantly outperforming existing state-of-the-art models such as MobileNetV2, VGG-16 with attention, and ResNet-50. These results demonstrate the model’s potential as a reliable and precise tool for breast cancer diagnostics, offering a robust framework for future applications in medical image analysis and contributing to improved clinical decision-making.
Keywords:	Accuracy, Breast cancer, Deep Learning, Sensitivity, Specificity, Thermograph
Source:	Journal of Theoretical and Applied Information Technology 15^th August 2025 -- Vol. 103. No. 15-- 2025
Full Text

Title:	RISK PREDICTION OF THEFT CRIMES IN URBAN COMMUNITIES: AN INTEGRATED MODEL OF LSTM AND ST-GCN
Author:	MEGAVATH TEJASHWINI , Dr M VENKATESWARA RAO , NAVNATH B. POKALE ,Dr N VENKATESH , NAVNATH KALE , Dr N ARJUN
Abstract:	Urbanization has transformed urban communities but also introduced challenges in managing crime and ensuring public safety. For prevention and control to be effective, a criminal risk prediction system is necessary. This paper presents a model that integrates Long Short-Term Memory Networks (LSTM) and Spatial-Temporal Graph Convolutional Networks (ST-GCN) to identify high-risk areas in cities. Using topological maps and crime data from Chicago, the model extracts spatial-temporal and temporal features to analyze theft patterns. Experimental results demonstrate its ability to predict crime occurrences within specific timeframes, offering a valuable tool for urban safety planning. The objectives of this study are (1) to develop an integrated LSTM + ST-GCN model for theft crime risk forecasting in urban communities, (2) to validate the model on real-world Chicago crime data, and (3) to benchmark its performance against standard baselines. Unlike existing methods that focus solely on spatial or temporal patterns, our hybrid approach fuses both dimensions for near real-time risk prediction, demonstrating up to 15% lower RMSE compared to standalone LSTM models.
Keywords:	Logistic Regression, Support Vector Machines (SVM), Naive Bayes, Predictive Models, Statistical Modeling.
Source:	Journal of Theoretical and Applied Information Technology 15^th August 2025 -- Vol. 103. No. 15-- 2025
Full Text

Title:	HYBRID EFFICIENTNET-TRANSFORMER ARCHITECTURE FOR AUTOMATED DETECTION OF GLAUCOMA
Author:	MALLA SIREESHA , MEKA JAMES STEPHEN , P.V.G.D. PRASAD REDDY
Abstract:	In recent years, the early and accurate diagnosis of Glaucoma, a leading cause of permanent blindness, has highlighted the importance of effective treatment and better patient care. Conventional deep learning methods that only use Convolutional Neural Networks (CNNs) have shown significant performance in classification tasks involving fundus images, but they often fail to capture global contextual dependencies that are crucial for fine-grained disease categorization. Addressing this limitation, a Hybrid CNN-Transformer architecture has been proposed that uses the local feature extraction capability of a pretrained EfficientNetB0 backbone and augments it with the global modeling power of Transformer encoders. The EfficientNetB0 module extracts robust spatial features, which are improved using Transformer layers to model inter-feature relationships throughout the image. The proposed Hybrid EfficientNet-Transformer model was trained and validated on a dataset created by combining five datasets containing Fundus images, employing standard preprocessing pipelines and label encoding strategies. Extensive experiments show that the proposed hybrid approach attains a 99% high classification accuracy on the unseen test set, outperforming many existing HybridCNNs and transfer learning models. These results indicate that the fusion of CNNs and Transformers offers a powerful framework for high-precision glaucoma detection, creating new prospects for the development of automated ophthalmic screening tools.
Keywords:	Glaucoma, Convolutional Neural Networks (CNNs), EfficientNetB0, Fundus images, Transformer encoders.
Source:	Journal of Theoretical and Applied Information Technology 15^th August 2025 -- Vol. 103. No. 15-- 2025
Full Text

Title:	DYSARTHRIC SPEECH RECOGNITION USING WAVENET CONVOLUTIONAL NEURAL NETWORK WITH PROSODY CONSISTENCY LOSS FUNCTION
Author:	ANEETA S ANTONY, ROHINI NAGAPADMA, AJISH K ABRAHAM
Abstract:	Dysarthric speech recognition concentrates on understanding speech impairments caused by neurological disorders. Speech recognition increases communication by enhancing adaptability and clarity for individuals. However, dysarthric speech recognition struggles to transcribe speech accurately due to variations in pitch, articulation, and rhythm that significantly differ from typical speech and processing dysarthric across different speakers. In this research, the WaveNet Convolutional Neural Network with Prosody Consistency Loss Function (WNCNN-PCLF) is proposed to recognise and classify dysarthric speech accurately. In CNN, WaveNet is incorporated for capturing local speech features that enhance the model’s ability to determine intricate variations and patterns in distorted speech. The PCLF assist in preserving natural speech patterns, which makes for more accurate rhythm and tone representation. Therefore, this integration enables better adaptation to dysarthric speech, which addresses both prosody and articulation issues effectively. Hence, the proposed WNCNN-PCLF achieves a high accuracy of 99.92% and 98.34% using UA-Speech and Kannada datasets compared to existing methods like Densely Squeezed and excitation Attention-gated Network (DySARNet).
Keywords:	Dysarthric Speech Recognition, Local Speech, Prosody Consistency Loss Function, Speakers, Wavenet Convolutional Neural Network.
Source:	Journal of Theoretical and Applied Information Technology 15^th August 2025 -- Vol. 103. No. 15-- 2025
Full Text

Title:	DEEP LEARNING-DRIVEN WEB INFORMATION EXTRACTION WITH CNN AND LSTM MODELS
Author:	B. BHAVANI , Dr. D.HARITHA
Abstract:	The high rate at which the amount of web data is growing has posed immense opportunities as well as challenges to the data extraction methods. Traditional web scraping solutions are easily outmatched by the dynamism and heterogeneity of web content leading to frequent failure and inefficiency. The article questions how these weaknesses could be mitigated by using adaptive deep learning models to extract web data by giving a robust and scalable solution to the problem. We propose an adaptive deep learning model learning and generalizing over diverse web structures and types of content using neural networks. It is a dynamic framework and can fit in changing web environments by utilizing domain adaptation and transfer learning to ensure consistency and accuracy in data extraction. We have determined through considerable experimentation that our adaptive architecture is considerably faster than the established scraping approach, with accuracy, structural change resilience and a reduction in manually configured dependency being noteworthy enhancements. These results illustrate the effectiveness of adaptive deep learning over web data extraction and lead to the path of much smarter and automated web scraping systems.
Keywords:	Web Data, Conventional Methods, Deep Learning, LSTM, Web-Scraping
Source:	Journal of Theoretical and Applied Information Technology 15^th August 2025 -- Vol. 103. No. 15-- 2025
Full Text

Title:	DIGITAL PLATFORMS FOR SUPPORTING ENTREPRENEURSHIP: EU EXPERIENCE FOR UKRAINE
Author:	Nataliia Tiahunova, Olena Zhuk, Viacheslav Kokhno, Pavlo Tereshchenko, Andrey Skrylnik
Abstract:	The aim of this research is to assess the impact of the use of digital platforms for entrepreneurship on business flexibility in the European Union (EU) and to provide recommendations for Ukraine. The study employed the following methods: correlation, regression, comparative, statistical analysis. The regression analysis found that digital technologies by themselves have a smaller impact on business flexibility than in combination with other important factors. These factors included talent, capital, regulation, technological infrastructure. The uniqueness of the Ukrainian resource Diia is noted, which integrates state services, services for businesses and citizens, and documents in one platform. A further direction for the platform’s development is proposed through the integration of AI-based technologies for intelligent form filling, automatic data verification, personalization of services, and provision of business advice. The results of the study may be useful for the development of digitalization policies at the state and enterprise levels by increasing the focus on the identified influencing factors.
Keywords:	Digital Platforms, Entrepreneurship, Artificial Intelligence, Capital, Regulation, Technological Infrastructure.
Source:	Journal of Theoretical and Applied Information Technology 15^th August 2025 -- Vol. 103. No. 15-- 2025
Full Text

Title:	APPLICATION OF ARTIFFICIAL INTELLIGENCE IN THE DEVELOPMENT OF EU STRATEGIES TO STRENGTHEN THE INSTITUTION OF POLITICAL ACCOUNTABILITY
Author:	YULIIA MELNYK, NADIIA HERBUT, INNA PIDBEREZNYKH, BOHDAN HRUSHETSKYI, MARINA SHULGA
Abstract:	Political accountability is a key element of democratic governance and transparency of state institutions. This is particularly relevant in the context of the implementation of EU standards, which imply increasing transparency, anti-corruption measures, and strengthening democratic principles. The study focuses on the analysis of the effectiveness of EU strategies in strengthening political accountability in member states and candidate countries, such as the Netherlands, Poland, Hungary, Ukraine, and Serbia. The importance of this study is determined by the need to develop effective mechanisms for adapting EU standards in different political contexts. The aim of the research is to study the impact of the implementation of European strategies on political accountability in selected countries, as well as identify the main factors that contribute to or hinder their effective implementation. The study employed the following methods: quantitative analysis to assess changes in political accountability metrics, sociological surveys to determine the level of trust in governments in five countries, comparative analysis to identify differences in the implementation of EU standards between member states and candidate countries, and correlation analysis to assess the relationship between the implementation of European standards and key indicators of political accountability. In addition, the research explores how artificial intelligence (AI) technologies can be applied in the development and monitoring of EU strategies, in particular through data-driven evaluation of public administration, risk modelling, and the identification of political misconduct patterns. The results of the study showed that EU member states demonstrate higher political accountability indicators due to the effective adaptation of European standards. Preliminary findings also suggest that AI-supported tools, such as algorithmic auditing and digital transparency platforms, can strengthen accountability mechanisms by enabling early detection of corruption risks and increasing civic oversight. The Netherlands consistently maintains a high level of democracy and transparency, while Poland and Hungary face certain challenges due to political polarization and centralization of power. Different trends are observed in candidate countries such as Ukraine and Serbia: Ukraine demonstrates gradual progress, while Serbia shows stagnation.
Keywords:	Political Responsibility, European Standards, Democracy, Government Transparency, Fight against Corruption, CPI, Social Reforms, Artificial Intelligence, Digital Governance, Algorithmic Accountability
Source:	Journal of Theoretical and Applied Information Technology 15^th August 2025 -- Vol. 103. No. 15-- 2025
Full Text

Title:	AN EFFECTIVE APPROACH FOR HISTOPATHOLOGIC ORAL CANCER PREDICTION USING BITTERLING FISH OPTIMIZATION TECHNIQUE
Author:	PULLAIAH PINNIKA , VENKATA RAO KASUKURTHI
Abstract:	Oral cancer poses a substantial global health concern, necessitating immediate action to mitigate its severe consequences. However, existing methods used for oral cancer classification exhibit limitations such as getting trapped in the local optima, high sensitivity to noisy data, and poor convergence speed, resulting in lower prediction reliability and clinical applicability. Hence, this research proposes a Bitterling Fish Optimization with Probability Entropy (BFO-PE) algorithm for feature selection during oral cancer classification. The PE enables the BFO to select optimal feature subsets by simulating the spawning behavior of bitterling fish, updating iteratively toward an optimal feature set, thereby enhancing performance. Initially, the Oral cancer (Tips and Tongue) Image (OCI) dataset and Kvasir Video and Image Repository (KVASIR) dataset are used to validate the model’s performance. The next step, pre-processing involves resizing images to standardized dimensions using interpolation. Then, feature extraction is performed using the ResNet 152 architecture, followed by which feature selection is carried out. Finally, the ResNet layer is deployed for classification of oral cancer into binary classes. The proposed approach attains a superior accuracy of 99.04% on OCI and 99.75% Kvasir datasets, outperforming existing methods like Convolutional Neural Networks - Improved Tunicate Swarm Algorithm (CNN-ITSA). These results demonstrate the effectiveness, generalizability, and potential clinical applicability of the developed system for oral cancer prediction.
Keywords:	Bitterling Fish Optimization; Convolutional Neural Networks; Improved Tunicate Swarm Algorithm; Oral Cancer And Probability Entropy.
Source:	Journal of Theoretical and Applied Information Technology 15^th August 2025 -- Vol. 103. No. 15-- 2025
Full Text

Title:	APPLICATION OF CHATGPT BY EDUCATORS AT THE TERTIARY LEVEL: SYSTEMATIC LITERATURE REVIEW
Author:	LIUDMYLA HOLUBNYCHA, OLENA KUZNETSOVA, NATALIA SOROKA, TETIANA SHCHOKINA, TETYANA KOSHECHKINA, OKSANA KOVALENKO
Abstract:	The advent of the conversational AI model, ChatGPT, has garnered considerable attention from educators, presenting substantial avenues for pedagogical innovation and holding extensive promise for integration by tertiary-level educators. However, a definitive understanding of optimal strategies for ChatGPT’s effective deployment within university pedagogy remains elusive. Consequently, this investigation undertakes a review of the extant scientific literature to address these critical inquiries. The purpose of the research is to thoroughly synthesize and critically evaluate the practical deployment of ChatGPT by tertiary level educators, involving a comprehensive assessment of its application across the initial two-year period following its introduction (November 30, 2022 – November 30, 2024) using a systematic review of scholarly publications. The study employed the PRISMA guidelines for data collection and evaluation, followed by a systematic review. The research findings derived from this process were then subjected to thematic analysis, organized around the identified themes and categories. The research results were examined based on the identified research questions. The results point to the determined in the article spheres of university educators’ activities referred to and considered in experimental scholarly publications, in which the ChatGPT use offers educationally valuable help and assistance to university educators. Pedagogical activities augmented by ChatGPT are critically examined, highlighting key findings, benefits and concerns the incorporation of ChatGPT may have in every determined sphere of tertiary level teaching. The recommendations to prevent or lessen the likely negative aspects of ChatGPT being part of university educators’ teaching practice are presented. The unique status of ChatGPT as a facilitative aid in university instructional settings is ascertained.
Keywords:	Application of ChatGPT, Higher education, Literature review, Teaching
Source:	Journal of Theoretical and Applied Information Technology 15^th August 2025 -- Vol. 103. No. 15-- 2025
Full Text

Title:	INFLUENCER CREDIBILITY AND BRAND REPUTATION IN SHAPING TRUST AND PURCHASE INTENTION: EXPLORING GENERATION Z'S RESPONSE TO LOCAL BEAUTY PRODUCTS ON TIKTOK
Author:	NATHASIA , HANSEN THENDY , JULIUS ROMARIO , ANISA LARASATI
Abstract:	Driven by societal pressures and beauty standards, skincare products have become a daily necessity, especially through TikTok, emerging as a prominent platform for beauty trends. This study explores the impact of brand reputation on purchase intention for local beauty products among Generation Z, examining the moderating role of TikTok influencer credibility and the mediating effect of trust. Although TikTok has become a powerful platform for shaping consumer behavior, particularly purchase intention, the extent to which influencer credibility moderates the relationship between brand reputation and purchase intention remains insufficiently examined. This research addresses this gap in the literature by proposing a single framework and conducting an experiment to assess how influencer credibility influences the strength of brand reputation’s effect on purchase intention. A quantitative approach was applied using a 2 (brand reputation: positive vs. negative) × 2 (influencer credibility: high vs. low) between-subjects experimental design, involving 135 participants who were randomly assigned to one of 4 different scenarios. The findings confirm that brand reputation significantly increases purchase intention. Moreover, an influencer with high credibility strengthens the positive impact of brand reputation on purchase intention. However, influencers with low credibility do not significantly affect consumer purchase decisions, highlighting the importance of authenticity and expertise over mere popularity. Interestingly, influencer credibility does not moderate the relationship between brand trust and purchase intention. Trust was found to fully mediate the relationship between brand reputation and purchase intention, serving as a crucial psychological mechanism that converts positive brand perceptions into buying behavior. This suggests that trust, grounded primarily in consumers’ direct brand experiences and overall reputation, is less influenced by external endorsements. These findings highlight the importance of managing brand reputation and partnering with credible influencers to maximize marketing effectiveness in Generation Z's purchasing decisions in the local beauty market.
Keywords:	Brand Reputation, Brand Trust, Purchase Intention, TikTok Influencer Credibility, Beauty Products
Source:	Journal of Theoretical and Applied Information Technology 15^th August 2025 -- Vol. 103. No. 15-- 2025
Full Text

Title:	MULTI-SCALE RETINEX-UNET FOR ENHANCEMENT OF LOW LIGHT WEAK CONTRAST IMAGES
Author:	MUDDAPU HARIKA, GOTTAPU SASIBHUSHANA RAO, RAJKUMAR GOSWAMI
Abstract:	The enhanced quality of images is crucial in the realm of image processing applications. However, images captured in low-light environments often suffer from poor contrast and noise, leading to a loss of detailed information. To address this challenge, we propose a Multi-Scale RETINEX-UNET (MSR-UNET) model for low-light weak contrast (LLWC) image enhancement. This novel approach integrates a modified U-Net architecture with an improved multi-scale Retinex (IMSR) model, aiming to preserve natural colors while enhancing visual quality. The proposed model is validated using the Renoir dataset, and its effectiveness is measured through PSNR (34.3 dB) and SSIM (99%). Compared to conventional enhancement techniques, our model outperforms existing methods by effectively reducing noise while maintaining structural details. This study advances deep learning-based image enhancement and sets a new benchmark for LLWC image processing.
Keywords:	Image Enhancement, Convolutional Neural Networks, Grey Level Co-occurrence Matrix, U-Net, Multi-Scale Retinex, Deep Learning.
Source:	Journal of Theoretical and Applied Information Technology 15^th August 2025 -- Vol. 103. No. 15-- 2025
Full Text

Title:	IMPLICATIONS OF USING ARTIFICIAL INTELLIGENCE TECHNIQUES TO ENHANCE A CHILD'S SOCIAL SKILLS
Author:	MHMMAD SULIEMAN AL SHAAR
Abstract:	Social skills are critical for a child’s development as they are fundamental for building relationships and succeeding academically and professionally. In comparison to the academic spheres of reading, language, and mathematics, these skills are just as valuable. However, children, particularly those with autism spectrum disorders (ASD) and social communication disorders, struggle with appropriate social behavior. With advancements in artificial intelligence (AI), such as natural language processing and computer vision, there are new ways to enhance social interactions for children. AI technologies can enable children to receive feedback in real time, engage them with social scenarios, and encourage social play through developmentally appropriate methods. There are, however, ethical, educational, and psychological concerns, such as data privacy, the implications of AI on child development, and the nature of interaction between machines and children. My goal in this study is to analyze how AI can be effectively utilized for the promotion of social skills in childhood, current applications and interventions, and point out gaps for future research. This research aims to draw attention not only to the existing models but also to the innovative approaches in order to highlight the potential of AI in transforming children’s development when used responsibly.
Keywords:	Social skill, Childhood, Artificial Intelligence, Social Interactions, School Performance
Source:	Journal of Theoretical and Applied Information Technology 15^th August 2025 -- Vol. 103. No. 15-- 2025
Full Text

Title:	DESIGN OF AN ITERATIVE METHOD FOR ENHANCED IOT DATA MANAGEMENT AND PROCESSING IN HEALTHCARE APPLICATIONS
Author:	VONTERU SRIKANTH REDDY, PADARTHI SWATHI, GOLLA USHA RANI
Abstract:	This study addresses the critical need for advanced frameworks in the realm of Internet of Things (IoT) data storage and processing, particularly within the healthcare sector. Traditional approaches often grapple with the complexity and diversity of IoT-generated data, leading to inefficiencies in data management and utilization. Existing methodologies fall short in accurately capturing the intricate relationships among heterogeneous data sources, including medical devices, patient information, and environmental parameters. Moreover, they struggle to process semantic aspects of data effectively, limiting the potential for nuanced analysis and interpretation operations. To address these gaps, this study introduces a novel IoT-ML (Internet of Things - Machine Learning) framework that integrates ontology-based data modeling, semantic Natural Language Processing (NLP), and adaptive Machine Learning (ML) for enhanced healthcare data processing. The proposed model includes a deep learning-based ontology generation engine, Transformer-based semantic interpretation (using BERT), and an adaptive, resource-aware ML schema to support real-time, energy-efficient data analysis.. A unified IoT-ML schema is also developed, integrating semantic and numeric data processing capabilities to support efficient data querying and interoperability across varied IoT platforms. The implementation of machine learning models introduces efficient data encoding strategies, reducing storage requirements while preserving data integrity levels. Adaptive learning algorithms are designed to accommodate the heterogeneity of IoT data, optimizing computational complexity. Furthermore, our models are resource-aware, dynamically adjusting to the computational and storage limitations of IoT devices & scenarios. Empirical evaluation using healthcare datasets from IEEE Data Port and Kaggle demonstrates the superiority of our proposed framework. Results reveal significant improvements over existing methods [4,5,16], including a 4.9% rise in precision, 4.5% improvement in accuracy, 5.9% rise in recall, 10.4% reduce in delay, 8.3% growth in AUC, and 5.5% improvement in specificity for classification tasks. These advancements underscore the potential of our proposed IoT-ML standard in revolutionizing healthcare data management, promising significant impacts on the precision, efficiency, and scalability of IoT data processing and storage sets.
Keywords:	IoT Data Management, Semantic Data Processing, Ontology-Based Modeling, Adaptive Learning Algorithms, Healthcare IoT Scenarios
Source:	Journal of Theoretical and Applied Information Technology 15^th August 2025 -- Vol. 103. No. 15-- 2025
Full Text

Title:	WHAT DRIVES IMPULSIVE BUYING IN VIDEO GAME MICROTRANSACTIONS? A STRUCTURAL EQUATION MODELING STUDY OF INDONESIAN GAMERS
Author:	FARZA ALIF MAHENDRA , RIYANTO JAYADI , TANTY OKTAVIA
Abstract:	Microtransactions in video games have become a growing monetization strategy but also trigger impulsive buying behavior. This study aims to evaluate the factors influencing impulse buying in video game microtransactions in Jabodetabek, focusing on key variables such as Shopping Enjoyment, Impulse Buying Tendency, Urge to Purchase, Positive Affect, Negative Affect, and Hedonic Motivation. This research employs a quantitative approach with 436 respondents, analyzed using Structural Equation Modeling (SEM) with SmartPLS 3.0. The findings indicate that Shopping Enjoyment enhances Positive Affect, which subsequently drives Urge to Purchase and Impulse Buying. Additionally, Impulse Buying Tendency significantly influences Impulse Buying, whereas Negative Affect and Hedonic Motivation show no significant impact. These findings provide insights for the gaming industry to develop more ethical monetization strategies and help players manage their purchasing behavior more effectively.
Keywords:	Impulsive Buying, Microtransactions, Monetization, Video Game, Structural Equation Modeling
Source:	Journal of Theoretical and Applied Information Technology 15^th August 2025 -- Vol. 103. No. 15-- 2025
Full Text

Title:	SECURE USER AUTHENTICATION AND KEY AGREEMENT IN SMART NETWORKS WITH BLOCKCHAIN GATEWAYS
Author:	DURVASI GUDIVADA , M. KAMESWARA RAO
Abstract:	As IoT-driven smart environments such as intelligent cities, buildings, and industries develop quickly, user authentication is becoming increasingly important to guard against illegal access and guarantee data security. Because of issues such as device heterogeneity, computing resource limitations, and potential single points of failure, conventional and current authentication methods frequently fail to overcome security issues such as impersonation, replaying, and denial-of-service attacks in IoT networks. With the goal of offering lightweight, secure, and decentralized user authentication and key agreement in smart settings, this study suggests a blockchain-based user authentication and secure key agreement protocol for smart networks. In this case, gateway nodes are permitted to create a blockchain to authenticate users and IoT nodes by integrating cryptographic hash algorithms. The AVISPA tool is used to formally verify a protocol’s security. The comparison analysis reveals that our protocol has an execution time of 0.0858 ms, confirming its performance against more recent research efforts.
Keywords:	IoT, User Authentication, Smart Contract, AVISPA
Source:	Journal of Theoretical and Applied Information Technology 15^th August 2025 -- Vol. 103. No. 15-- 2025
Full Text

Title:	MINERAL IDENTIFICATION USING ENSEMBLE OF HANDCRAFTED AND DEEP LEARNING FEATURES WITH XGBOOST
Author:	SIRAM DIVYA , KUNJAM NAGESWARA RAO
Abstract:	The industry that utilizes beach sand minerals containing titanium zirconium and other strategic elements requires precise classification of mineral grains. The identification techniques commonly used today take long periods to complete and need experts to interpret them while producing results that lack consistency. A well-annotated MINET (Mineral Identification NETwork) dataset of high-resolution mineral grain images serves this study to develop a powerful automated classification machine learning pipeline which addresses previous limitations. The proposed framework uses both handcrafted texture color and shape features with deep ResNet50 model features from extracted representations. These joint representations produce an advanced system which improves the identification of intricate mineral formations. Our framework uses XGBOOST as its classifier to show how features drawn from both handcrafted and deep learning extraction boost automated petrography systems and strengthens MINET's position as a critical benchmark for mineral recognition intelligence.
Keywords:	Mineral Gains,MINET,Handcrafted Features,Resnet50,XGBOOST
Source:	Journal of Theoretical and Applied Information Technology 15^th August 2025 -- Vol. 103. No. 15-- 2025
Full Text

Title:	HYBRID ATTENTION-BASED DEEP LEARNING MODEL FOR WIND SPEED FORECASTING IN RENEWABLE ENERGY APPLICATIONS
Author:	HEMANTH SAI MADUPU, BALA SAIBABU BOMMIDI, IMRAN ABDUL, K. RAMBABU, BEJJAM S N BENARJI
Abstract:	Accurate wind speed forecasting (WSF) is critical for the efficient integration of wind energy into power systems. In this study, a novel hybrid model combining Ensemble Empirical Mode Decomposition (EEMD), Convolutional Neural Network (CNN), and Attention-based Long Short-Term Memory (ALSTM)—referred to as EEMD-CNN-ALSTM—is proposed for short-term wind speed prediction. The model leverages EEMD to decompose complex and non-stationary wind speed signals into more manageable components, CNN to extract localized spatial features, and ALSTM to selectively capture temporal dependencies using attention mechanisms. The proposed model is evaluated using 1-hour interval wind speed data collected from two distinct wind farms: Garden City and Idalia. Performance is assessed using standard metrics including RMSE, MAE, MSE, and R². Results demonstrate that the EEMD-CNN-ALSTM model consistently outperforms benchmark models and other hybrid architectures. Across both wind farms, the proposed model achieves the lowest error rates and highest predictive accuracy, confirming its robustness and effectiveness in handling the complexities of WSF.
Keywords:	Attention-Based LSTM, Hybrid Model, Time Series Prediction, Deep Learning, Renewable Energy, Short-Term Forecasting
Source:	Journal of Theoretical and Applied Information Technology 15^th August 2025 -- Vol. 103. No. 15-- 2025
Full Text

Title:	ROUGH SET BASED PRIVACY PRESERVATION OF HEALTHCARE DATA USING ASSOCIATION RULE MINING AND GENETIC ALGORITHM
Author:	SRUTIPRAGYAN SWAIN, BIBHUTI BHUSAN DASH, SUNEETA MOHANTY, BANCHHANIDHI DASH, PRASANT KUMAR PATTNAIK
Abstract:	Smart healthcare refers to the combination of emerging trends like Big Data Analytics, IoT and AI & ML. The shift from traditional systems to smart health applications results in the generation of massive volumes of data that need to be managed. This data must be stored on cloud servers and shared with organizations for future research and applications. However, sharing it with third-party servers can result in the exposure of sensitive patient information to external entities. A significant challenge in this context is ensuring the protection of sensitive information to maintain privacy. Consequently, attribute reduction plays a crucial role in managing large datasets by removing unnecessary or redundant data, thereby facilitating the efficient hiding of sensitive rules before public disclosure. This paper presents a privacy-preserving framework designed to hide sensitive fuzzy association rules. The proposed model incorporates two key stages: a pre-activity phase for mining fuzzified association rules and a post-activity phase for concealing sensitive rules. Experimental findings confirm the effectiveness of the proposed approach.
Keywords:	Rough Set (RS), Fuzzy Proximity Ratio (FPR), OIS (Ordered Information System), Fuzzy Association Rule Mining.
Source:	Journal of Theoretical and Applied Information Technology 15^th August 2025 -- Vol. 103. No. 15-- 2025
Full Text

Title:	INTEGRATING BIG DATA ANALYTICS INTO HEALTHCARE DECISION SUPPORT SYSTEMS FOR BETTER PATIENT OUTCOMES
Author:	ATLURI LAKSHMI , TADAVARTHI NAGA NAVYA , NELAKUDITI KRISHNAVENI , NELLI SREEVIDYA , D. NAGA PURNIMA , T.P.S. KUMAR KUSUMANCHI
Abstract:	This article investigates the incorporation of BDA with healthcare DSSs to improve the quality of clinical decision-making and patient outcomes. This study aims to assess the potential of BDA in analyzing a broad range of healthcare data types, from EHRs to real-time sensor data and clinical histories, to aid in the prognostication of patient conditions, customized treatment, and Candice optimization. The approach deploys machine learning classifiers (ANN, RF, and SVM) on a synthetic dataset of over 500 thousand patient records. The findings show that the performance of the ANN model is better than the conventional machine learning models with an accuracy of 91.6%, precision of 0.89, recall of 0.87, and AUC-ROC of 0.92. Comparison with the current DSS models shows superior predictive precision, decision efficiency, and real-time intervention capacity. This incorporation of big data improves diagnostic and therapeutic results but also allows physicians to make real-time suggestions when making decisions, thereby allowing early clinical interventions. Significance This research contributes to realizing the transformative capabilities of BDA in the modernization of healthcare, the reduction of operation ineffectiveness, and patient services with a substantial impact.
Keywords:	Big Data Analytics, Healthcare Decision Support Systems, Machine Learning, Predictive Modeling, Patient Outcomes, Artificial Neural Networks
Source:	Journal of Theoretical and Applied Information Technology 15^th August 2025 -- Vol. 103. No. 15-- 2025
Full Text

Title:	EARLY FIRE HAZARD PREDICTION FRAMEWORK IN SMART CITIES USING DEEP LEARNING WITH ANTLION OPTIMIZATION ALGORITHM
Author:	DR.G. BHUVANESWARI, DR.G. MANIKANDAN, M. SANDHYA3, DHANESH KUMAR, DR. ZIAUL HAQUE CHOUDHURY, PRABHAKARA RAO T
Abstract:	This study aims to refine the early fire risk prediction model for evaluating the accurate locations of fire using sensor data. Internet of Things (IoT) is integral to smart cities. IoT applications in smart cities include crime predictions, traffic optimization and monitoring of health and environmental conditions. This article reports a study on using Recurrent Neural Network (RNN) with Ant Lion optimization (ALO) framework to enhance and refine the prediction of fire hazards. IoT sensors in smart cities monitor the environmental conditions such as drought, temperature, smoke, flame, relative humidity, fuel moisture and duff moisture. This sensed data is stored in the firebase cloud storage and analyzed in the MATLAB tool. The ensuing enhancement of the proposed model is validated by comparison with conventional prediction models. Our results indicate gains in accuracy and reduction in error rates in fire hazard predictions.
Keywords:	Environment, Fire Hazards, Internet Of Things, Smart City.
Source:	Journal of Theoretical and Applied Information Technology 15^th August 2025 -- Vol. 103. No. 15-- 2025
Full Text

Title:	TASK OFFLOADING IN EDGECLOUD RESOURCE SCHEDULING USING DEPENDANT COMPUTATION TASK OFFLOAD USING ENHANCED FIREFLY (DCTO-EFF)
Author:	K.VINOTHKUMAR, DR. D. MARUTHANAYAGAM
Abstract:	In order to efficiently distribute computing jobs between edge devices and cloud resources, task offloading is a crucial strategy in edge-cloud resource scheduling. Numerous optimization strategies have been put forth to maximize system performance and resource utilization. Among the cutting-edge optimization algorithms that are carefully investigated in this study are the Dependant Computation Task Offload Using Enhanced Bacterial Foraging Optimization (DCTO-EBFO), Firefly Algorithm (FA), Elephant Herding Optimization Algorithm (EHOA), Social Spider Optimization (SSO), Bee Colony Optimization (BCO), and Hybrid Grey Wolf Lion Optimization (HGWLO). However, most existing optimization strategies fail to effectively handle dependent computational tasks and often suffer from slow convergence and reduced efficiency under dynamic edge-cloud conditions. Furthermore, Proposed Dependant Computation Task Offload utilizing Enhanced Firefly (DCTO-EFF), a novel optimization method, is presented and assessed. At the beginning of the paper, the importance of optimization strategies in enhancing offloading alternatives is discussed, along with an overview of job offloading in edge-cloud resource scheduling. Next, a detailed explanation of each optimization algorithm's benefits and core concepts is provided. The suggested DCTO-EFF algorithm for work offloading is then explained, along with its features, capabilities, and advantages. Comprehensive simulations using EdgeCloudSim are used to compare the efficacy of the optimization strategies. Performance measurements include things like makespan, reaction time, energy consumption, resource utilization, latency, convergence speed, execution time, and delay. All things considered, this study provides useful details regarding the advantages and disadvantages of various optimization algorithms for task offloading in edge-cloud resource scheduling, assisting practitioners and researchers in selecting the optimal algorithm based on specific deployment scenarios and application needs.
Keywords:	Task Offloading, Edge-Cloud Resource Scheduling, Optimization Algorithms, Firefly Algorithm, Elephant Herding Optimization Algorithm, Dependant Computation Task Offload using Enhanced Firefly.
Source:	Journal of Theoretical and Applied Information Technology 15^th August 2025 -- Vol. 103. No. 15-- 2025
Full Text

Title:	TWO LEVEL DENOISING WITH IMAGE BOUNDARY PIXEL BASED SEGMENTATION OF LUNG MRI FOR ACCURATE LUNG TUMOR DETECTION
Author:	ANJANEYULU GURRAM , PARTHASARATHY RAMADASS
Abstract:	Cancer is characterized by abnormal clusters of cells and manifests itself in a variety of ways. In India, lung cancer is the second leading killer as per the records of Indian Cancer Society. Patients have a far better chance of survival when they undergo early detection procedures using Magnetic Resonance Imaging (MRI) scans. The ability to segment MRI scans is crucial for accurate diagnosis and treatment in the clinic, making this an essential and important task. Nevertheless, flaws like low contrast, noise, intensity inhomogeneity, etc., frequently taint MRI images. Clinical diagnostics and medical research rely heavily on MRI, however noise interference frequently compromises the imaging process. This noise comes from all over, and it lowers the image quality, which makes it harder for doctors to make correct diagnoses based on the information they see. Conventional denoising techniques fail to account for MRI images' more intricate forms of noise, such as Rician noise, because they presume that noise is normally distributed normally. Consequently, denoising is still an important and difficult task. Inhomogeneous intensities, juxta-pleural nodules, image noises, and other similar issues make accurate lung segmentation from medical images a continuing problem. An efficient noise-denoising approach for images should keep critical edges intact. There have been several attempts to eliminate noise using image denoising algorithms, however it is often not possible to do so completely. But it takes a long time, is easy to make mistakes with, and needs medical knowledge. This research aims to increase the accuracy of lung area segmentation in MRI images by proposing an updated two-level denoising and segmentation framework for lung region segmentation preserving the accurate lung border. To denoise MRI images of the lungs while keeping their outlines intact, the proposed system initially employs a filtering strategy based on image decomposition. The MRI images of the lungs are subsequently segmented using a combination of enhanced morphological techniques. By utilizing a contour correction strategy that is based on a quick corner detection technique, the segmentations are further enhanced in order to rectify and smooth the retrieved lung outlines. This research proposes a Two Level Denoising with Image Boundary Pixel based Segmentation (TLD-IBPbS) model for accurate lung tumor detection. The proposed model is compared with the traditional models and the results represent that the proposed model performance is high.
Keywords:	Lung Cancer, Magnetic Resonance Imaging, Denoising Techniques, Lung Segmentation, Two-Level Denoising, Enhanced Morphological Techniques, Image Boundary Pixel, Lung Tumor Detection.
Source:	Journal of Theoretical and Applied Information Technology 15^th August 2025 -- Vol. 103. No. 15-- 2025
Full Text

Title:	A UNIFIED CNN-BASED FRAMEWORK FOR GENERALIZED AND REAL-TIME COLORECTAL CANCER PREDICTION AND DIAGNOSIS: BRIDGING DATA GAPS, ENHANCING INTERPRETABILITY, AND PERSONALIZING OUTCOMES
Author:	K YOGESWARA RAO , S ADI NARAYANA
Abstract:	Colorectal cancer (CRC) remains one of the most common and deadly cancers worldwide, making early detection and accurate diagnosis more important than ever. In this work, we introduce UniCRC-Net—a smart, CNN-based system designed to predict and diagnose colorectal cancer in real time, using structured patient data. Unlike many current machine learning and deep learning models, which struggle with scattered data, lack of explainability, and generic predictions, this unified approach brings together multiple patient details—like age, gender, pathology scores, gene markers, diet, and environment—into a streamlined and intelligent framework. The model is trained on a carefully constructed synthetic dataset and optimized using the Adam algorithm over 50 training epochs. It performs exceptionally well, hitting a perfect 100% accuracy, F1-score, and AUC, which means it’s both highly precise and consistent in identifying cancer cases. The results are backed by clear visualizations—such as accuracy and loss graphs, a confusion matrix, and a sharp ROC curve—demonstrating how stable and dependable the model is throughout its training. What sets UniCRC-Net apart is its real-time capability, its ability to personalize predictions, and its transparent design, which makes it easier to trust in clinical use. It's also built with the future in mind—ready for integration with federated learning systems that protect patient privacy while enabling collaboration across hospitals and regions. In short, this framework not only fills major gaps in CRC diagnostics but also moves us a step closer to AI-powered, patient- specific cancer care that’s fast, secure, and clinically meaningful.
Keywords:	Colorectal Cancer, Deep Learning, Convolutional Neural Network, Real-Time Diagnosis, Personalized Healthcare, ROC-AUC, Clinical Decision Support and Interpretability in AI
Source:	Journal of Theoretical and Applied Information Technology 15^th August 2025 -- Vol. 103. No. 15-- 2025
Full Text

Title:	EFFICIENT BIG DATA STORAGE SOLUTIONS FOR DISTRIBUTED CLOUD COMPUTING SYSTEMS
Author:	RAVURI DANIEL, BODE PRASAD, Y SREERAMAN, KANDRAKUNTA CHINNAIAH, DORABABU SUDARSA, INDHUMATHI RAVICHANDRAN, MANIDHEER BABU GORIKAPUDI
Abstract:	Storing large volumes of data in distributed cloud computing systems can cause a problem, as managing both kinds of data, structured or unstructured, is difficult with existing tools like the Hadoop Distributed File System (HDFS) and object storage systems. Existing solutions are not effective at managing both throughput, growth capacity, and affordability simultaneously, particularly for data workloads with a wide range of use patterns. Although HDFS is designed for ordered data and Amazon S3 handles a large amount of unstructured data, there are no solutions in these approaches to quickly access or allocate resources to different kinds of data. To address this gap, a new hybrid architecture is proposed that combines HDFS with Amazon S3 and allocates data according to its type and how often it is used. The aim of this research is to explore how it integrates different systems to improve data retrieval time, ensure fault tolerance, and make everything cost-efficient for both old and new data in the cloud. By measuring the systems using the KDD Cup 1999 and UNSW-NB15 datasets, it is found that the hybrid model retrieves data faster, more reliably, more scalable, and more economical than HDFS and object storage used alone. The results of this study describe a stable and practical solution for data storage in the cloud and set a path for new studies and practical cloud data storage improvements.
Keywords:	Hybrid Storage, Big Data, Distributed File Systems, Object Storage, Data Retrieval, Cloud Computing
Source:	Journal of Theoretical and Applied Information Technology 15^th August 2025 -- Vol. 103. No. 15-- 2025
Full Text

Title:	NEURONET: AN ADVANCED DEEP LEARNING FRAMEWORK FOR EARLY DIAGNOSIS AND PREEMPTIVE TREATMENT OF NEURODEGENERATIVE DISEASES
Author:	ASAD HUSSAIN SYED , AYESHA SIDDIQUA , K. MOHAMMADI JABEEN , SABIHA MAHVEEN
Abstract:	Globally, citizens are bearing a heavy burden about the ravages of neurodegenerative diseases such as Alzheimer’s and Parkinson’s. These diseases develop slowly and are frequently diagnosed at a late stage when therapy possibilities are restricted. Currently level of diagnostics rely on clinical examination and imaging but many times early pathological clues are overlooked. AI driven decay prediction models, including Capsule Networks (CapsNets) and Sparse Learning Models are developed and achieve improvement, but not efficient enough to fuse multimodal data jointly and obtain spatial-temporal features which are critical for early detection and prediction. To fill these gaps, we propose a novel hybrid deep learning framework named NeuroNet, which is composed of 3D Convolutional Neural Networks (CNN) for learning spatial features, Long Short-Term Memory (LSTM) networks for modeling temporal sequences, attention mechanism for selecting diagnostic regions and clinical biomarkers for multimodal information fusion. This architecture is fine-tuned by Bayesian hyperparameter optimization, and validated through ADNI dataset. The primary contribution of this work is the evidence that multimodal fusion with attention-informed deep learning, drives a dramatic improvement in diagnostic accuracy and interpretability. Accuracy, precision and recall achieved by NeuroNet were 98.62%, 98.50%, 98.70%, and an AUC–ROC value of 0.998, respectively, showing a superior performance over several state-of-the-art models. Explainable AI techniques like Grad-CAM and SHAP increase model transparency even more. This work provides new insights in the form of a clinically viable and interpretable deep learning model which pushes the frontier of AI-facilitated neurodegenerative disease diagnosis. The model can be easily incorporated in clinical diagnostical systems for early intervention and better prognosis.
Keywords:	NeuroNet, Deep Learning, Neurodegenerative Diseases, Multimodal Integration, Early Diagnosis
Source:	Journal of Theoretical and Applied Information Technology 15^th August 2025 -- Vol. 103. No. 15-- 2025
Full Text

Title:	DEEP LEARNING-POWERED SKIN DISEASE CLASSIFICATION: OPTIMIZING TRANSFER LEARNING FOR IMPROVED ACCURACY
Author:	SIPRA SAHOO , PRABHAT KUMAR SAHU, SMITA RATH, MITRABINDA KHUNTIA , MONALISA PANDA , DEEPAK KUMAR PATEL , NIBEDITA JAGADEV , SHRABANEE SWAGATIKA
Abstract:	Skin diseases represent a prevalent global health challenge, with prevalence rates in India ranging from 7.9% to 60%. While deep learning approaches show promise for automated diagnosis, existing methods exhibit limitations in robustness and generalizability due to: (1) inadequate feature selection methodologies, (2) suboptimal data augmentation strategies, (3) unexplored hybrid frameworks and (4) insufficient validation of attention mechanisms in clinical settings. These study addresses these gaps through three key innovations: comprehensive optimizer-architecture benchmarking, a novel dual-dataset validation framework and an integrated preprocessing pipeline combining seven enhancement techniques. We systematically evaluate four deep learning architectures (custom CNN, VGG16, DenseNet-121, Inception-ResNet-v2) with three optimization algorithms (Adam, SGD, RMSprop) on partitioned HAM10000 datasets containing 10,015 dermatoscopic images across seven disease categories. Our approach reveals new knowledge: (1) VGG16 with Adam achieves state-of-the-art 93.14% accuracy- the highest reported for single-model HAM10000 classification; (2) RMSprop unexpectedly outperforms Adam for DenseNet121 (83.86% vs 81.43%); and (3) dataset-specific optimizer behaviors critically impact clinical applicability. These findings establish that systematic evaluation of the model optimizer data set significantly improves diagnostic robustness. Research provides a foundation for affordable and accessible diagnostic tools with clinically actionable insights for deployment optimization, which could benefit populations with limited healthcare access.
Keywords:	Deep Learning, Classification, Convolutional Neural Network, VGG16, DenseNet-121, Inception-ResNet-v2, Transfer Learnig, Fine-Tuning.
Source:	Journal of Theoretical and Applied Information Technology 15^th August 2025 -- Vol. 103. No. 15-- 2025
Full Text

Title:	KNOWLEDGE-INFUSED TRANSFORMER FOR ASPECT-BASED SENTIMENT ANALYSIS USING CONCEPTNET AND SENTICNET WITH CROSS-DOMAIN GENERALIZATION
Author:	RAMESH BABU PITTALA, RAMAKRISHNA BOMMA, VEERASWAMY PITTALA, MEDIKONDA ASHA KIRAN, GARLAPATI NARAYANA, SOLLETI PHANI KUMAR, MANYAM THAILE, LAKSHMI PRASANNA BYRAPUNENI
Abstract:	This paper introduces Knowledge-infused Transformer Aspect-based System, a novel approach to address challenges in Knowledge-Infused Transformer for Aspect-Based Sentiment Analysis Using ConceptNet and SenticNet with Cross-Domain Generalization. The KTAS framework is designed with powerful algorithms that enable it to increase performance metrics by slightly more than 29 percent improvement over current practices. The feasibility of the framework is evident, as demonstrated by the accuracy of the results in an experiment using standard datasets. Specifically, the suggested system involves the combination of different computer-based methods such as group theory, dynamical systems, and federated learning to provide a resistant design that would excel the existing state-of-the-art approaches. KTAS is shown to attain better performance than others in various evaluation aspects, as evidenced in a detailed testing on Penn Treebank data and MS COCO data, on the results of which the current work is based. The implementation of the above limitations in current solutions is to take advantage of multimodal fusion and interpretability mechanisms, and therefore better manage the complex data patterns. The experiment's results indicate that the methodology applies to real-world scenarios with limited resources and computational complexity, as it is highly accurate. There are also ablation studies done to compare the input made by each component to the entire performance, and it can be seen that the encoder-decoder module is especially exceptional in ensuring the best results. Furthermore, a sensitivity analysis is conducted, which determines the risks regarding the robustness of KTAS in different situations, as well as showing that it is stable in various working conditions. The theoretical analysis gives formal requirements for the convergence character and computational effectiveness of the algorithm. Lastly, they discuss the possible areas of the approach and also provide a guideline towards future research that can be done to extend the capacities of the proposed system. There is substantial validation to support confidence in the proposed system.
Keywords:	Cybersecurity, Human-Computer Interaction, Natural Language Processing, Data Mining, Software Engineering
Source:	Journal of Theoretical and Applied Information Technology 15^th August 2025 -- Vol. 103. No. 15-- 2025
Full Text

Title:	FORECASTING CURRENCY EXCHANGE RATES USING ARIMA, ETS AND RNN: A MACHINE LEARNING PERSPECTIVE
Author:	T. SONI MADHULATHA , DR. MD. ATHEEQ SULTAN GHORI
Abstract:	Time series forecasting plays a key role in financial decision-making, particularly in predicting exchange rate fluctuations. In this paper we implemented three different methods for forecasting the INR/USD exchange rate: The Autoregressive Integrated Moving Average (ARIMA) method, the Exponential Smoothing (ETS) method, and a Long Short-Term Memory (LSTM)-based Recurrent Neural Network (RNN) approach. The ARIMA (5, 1, and 0) model captures past dependencies using an autoregressive framework, while the ETS model adjusts for trend-based variations using exponential smoothing techniques. The LSTM-based RNN, designed to capture non linear patterns, is trained using MinMax-scaled data with a sequence-based structure. When comparing all 3 models in terms of predictive accuracy, the deep learning models capturing short-term fluctuations and nonlinear trends more effectively. And performance evaluation using Mean Squared Error (MSE) as 36.424298, 17.956008, 0.521610, Mean Absolute Error (MAE) 5.333123, 3.766352, 0.49238, and Root Mean Squared Error (RMSE) as 6.035255, 4.237453, and 0.64872confirms the superior forecasting capability of the deep learning-based approach. These findings highlight the potential of RNNs in financial time series forecasting, offering a robust alternative to traditional statistical models.
Keywords:	Currency Forecasting, RNN, ARIMA, ETS, Machine Learning
Source:	Journal of Theoretical and Applied Information Technology 15^th August 2025 -- Vol. 103. No. 15-- 2025
Full Text

Title:	HOW CAN GIS INTEGRATION IMPROVE LAND MANAGEMENT FOR SUSTAINABLE DEVELOPMENT IN TERRITORIAL COMMUNITIES
Author:	SERHII SHEVCHUK , NATALIIA PROKOPENKO , ANDRIY VYNNYK , OLEKSII MUKHIN , DENYS SITKO
Abstract:	Despite the increasing availability of open geospatial data and the widespread use of geographic information systems (GIS), there remains a significant gap in the literature regarding the systematic integration of GIS into community-level land management for sustainable development. Existing studies often focus on isolated applications or sectoral use cases, lacking a holistic framework for operational implementation in decentralised settings. This study addresses this gap by developing and validating a comprehensive concept of GIS integration into land management processes, tailored to the needs of territorial communities. The proposed approach covers the full analytical cycle: from the collection, cleaning, and normalisation of satellite and cadastral data to environmental sustainability assessment, spatial visualisation, and decision-making. Drawing on multi-source data (Sentinel-2, Landsat 8, public cadastral registers, crowdsourcing platforms), and using QGIS, ArcGIS, GeoDA, and Google Earth Engine, the study conducts a comparative analysis of software capabilities in spatial modelling, including the construction of NDVI, LST, and ESI indices, interpolation, regression, and scenario-based cluster analysis. A novel Environmental Sustainability Index (ESI) is introduced, integrating factors such as urbanisation, vegetation cover, erosion susceptibility, and anthropogenic pressure, and visualised as multilayer thematic maps. A metropolitan case study demonstrates the operational value of GIS in land use planning and provides an adaptable algorithm for community-level implementation. The results contribute new knowledge by establishing a replicable GIS integration model that enhances spatial accuracy, planning transparency, and environmental resilience. Key challenges identified include standardising data formats, expanding digital skills, and aligning with national legislation. This study offers both methodological innovation and practical tools for advancing sustainable land governance in resource-constrained environments.
Keywords:	GIS; Sustainable Development; Territorial Communities; Environmental Sustainability; Natural Resources
Source:	Journal of Theoretical and Applied Information Technology 15^th August 2025 -- Vol. 103. No. 15-- 2025
Full Text

Title:	MODEL AGNOSTIC META LEARNING – LONG SHORT-TERM MEMORY FOR LEARNING STYLE CLASSIFICATION
Author:	AMRUTH K JOHN , BINU THOMAS
Abstract:	An adaptive learning system aims to enhance the effectiveness of the educational process by tailoring it to individual students. A key aspect of this adaptation involves identifying the most suitable learning approach, based on Visual, Auditory, and Kinesthetic (VAK) learning styles. However, accurately classifying learning styles remains a challenge due to the presence of concept drift, which affects the network’s ability to generalize across different learners. To address this, the Model Agnostic Meta Learning – Long Short-Term Memory (MAML-LSTM) model is proposed in this research study for effective learning style classification. MAML is incorporated into the LSTM network to identify shifts in classification patterns and adapt to new learners quickly. Rather than retraining the network from the beginning, the model dynamically fine-tunes the LSTM in response to concept drift, thereby improving its generalization capability. The MAML-LSTM integration enables rapid adaptation to concept drift by fine-tuning on limited new data, eliminating the need for complete retraining. This enhances the model’s ability to maintain high classification accuracy across dynamic learner behaviors. Additionally, Local Interpretable Model-agnostic Explanations (LIME) are employed after classification to highlight key features, ensuring greater transparency and interpretability. The proposed MAML-LSTM achieves 97.77% accuracy, 97.72% precision, 97.72% recall, 97.72% F1-score, 97.72% specificity, and 99.81% AUC on the VAK learning style dataset, outperforming existing algorithms.
Keywords:	Auditory, Kinesthetic, Learning style, Long Short-Term Memory, Model Agnostic Meta Learning and Visual.
Source:	Journal of Theoretical and Applied Information Technology 15^th August 2025 -- Vol. 103. No. 15-- 2025
Full Text

Title:	EARLY SEPSIS DETECTION FROM ICU DATA USING WAVELET FEATURE EXTRACTION WITH HYBRID DEEP LEARNING FRAMEWORK
Author:	Dr. V. GOKULA KRISHNAN , Dr. PINAGADI VENKATESWARA RAO , Dr. K. SREERAMAMURTHY , Dr. R. DHARMAPRAKASH , Dr. B. PRATHUSHA LAXMI
Abstract:	For patients in intensive care units (ICUs) to have better outcomes after suffering from sepsis, a potentially fatal illness, prompt and precise identification is essential. Timely care and better patient outcomes depend on early sepsis detection. It is common for traditional diagnostic procedures to miss the early signs of sepsis, which means that treatment is delayed. Using the open-source MIMIC-III clinical database, we provide a new hybrid diagnostic system for early sepsis identification that makes use of deep learning, signal processing, and metaheuristic optimisation. We start by using physiological time-series data including temperature, respiration rate, blood pressure, and heart rate to extract robust, multi-scale, and noise-resistant features using the Wavelet Scattering Transform (WST). Afterwards, a one-dimensional input-adapted AlexNet-based Convolutional Neural Network is trained with these features to learn the high-level feature representations necessary to differentiate among septic and non-septic patients. We optimise hyper parameters including learning rate, filter sizes, and dropout rates using the Hunger Games Search (HGS) method, a new metaheuristic inspired by nature, to further enhance model accuracy and generalisation. Benchmark CNN architectures and traditional machine learning models are outperformed by the suggested WST-AlexNet-HGS pipeline. Classification performance above 95% as shown by evaluation measures like accuracy, precision, recall, and F1-score, demonstrating strong diagnostic reliability. Given the model's efficacy, it might be used as a clinical decision support tool in intensive care units operating in real-time. This hybrid pipeline demonstrates >95% accuracy in early detection of sepsis from ICU data, outperforming conventional methods. The model's robustness and reliability position it as a viable decision support system for real-time clinical use.
Keywords:	Intensive care units, Hunger Games Search, Wavelet Scattering Transform, Sepsis detection, Convolutional Neural Network, AlexNet.
Source:	Journal of Theoretical and Applied Information Technology 15^th August 2025 -- Vol. 103. No. 15-- 2025
Full Text

Title:	SWARM OPTIMIZED MACHINE LEARNING MODEL FOR ENHANCED PREDICTION OF CORONARY ARTERY DISEASE
Author:	JAYAMOL P. JAMES , Dr. GNANAPRIYA S.
Abstract:	Coronary artery disease poses a critical health challenge requiring accurate and timely diagnosis. This research introduces an optimized classification framework that integrates Artificial Bee Colony (ABC) algorithm with Naive Bayes (NB) to strengthen prediction performance. The working mechanism begins with feature extraction from a benchmark coronary dataset, where ABC acts as a bio-inspired swarm intelligence method to identify the most relevant clinical attributes by mimicking the intelligent foraging behavior of bees. By eliminating redundant and noisy features, ABC enhances the learning environment for the Naive Bayes classifier. The probabilistic nature of NB then utilizes these refined features to compute posterior probabilities for classification, ensuring efficient decision boundaries. Experimental validation shows that the ABC-NB model achieves 93.45% accuracy, 94.23% sensitivity, and 92.56% specificity, outperforming standard classifiers in early-stage CAD detection. The model’s simplicity, interpretability, and high predictive value make it suitable for integration into clinical workflows. Its low computational complexity enables deployment on embedded medical systems. This framework promotes scalable, cost-effective, and accurate cardiac risk prediction. Future advancements will involve real-time prediction support, dynamic retraining with streaming patient data, and expansion to multi-modal datasets for comprehensive cardiovascular profiling across diverse populations and clinical scenarios.
Keywords:	Coronary Artery Disease, Artificial Bee Colony, Naive Bayes, Feature Selection Medical Diagnosis, Swarm Intelligence
Source:	Journal of Theoretical and Applied Information Technology 15^th August 2025 -- Vol. 103. No. 15-- 2025
Full Text

Title:	DIGITAL TRANSFORMATION OF PUBLIC SERVICES SYSTEMATIC REVIEW OF KEY SUCCESS FACTORS
Author:	MOUTAMIR Hajar , El QOUR Tahar
Abstract:	The digital transformation of public services, particularly through e-government, is crucial to increasing the efficiency, transparency and accessibility of government services, especially in developing countries. Although significant efforts have been made, public sector organizations continue to face persistent challenges in the implementation of digital services, often leading to suboptimal outcomes. While numerous studies have explored critical success factors (CSFs) across different contexts, a systematic consolidation of these findings remains limited, hindering both theoretical integration and practical application. This systematic review addresses that gap by applying the PRISMA approach to data collected between 2004 and 2024 from the Scopus and Web of Science databases, aggregating findings from multiple case studies conducted across various countries. The analysis consists in identifying and detailing the determining factors for the successful digital transformation of public services, in order to support the implementation of new initiatives. The results highlight factors such as strong leadership and governance, availability of technological infrastructure, security and confidentiality of systems, as well as quality of systems, information, services and ease of use. In conclusion, this study proposes a structured framework for the implementation of sustainable and scalable digital public services, and enhancing the inclusiveness of public services.
Keywords:	Key Success Factors, Determinants, Digital Transformation, Public Services, E-Government.
Source:	Journal of Theoretical and Applied Information Technology 15^th August 2025 -- Vol. 103. No. 15-- 2025
Full Text

Title:	UTILIZING BIG DATA FOR SOCIAL MEDIA TREND FORECASTING AND INFLUENCE ANALYSIS
Author:	SHOBANA GORINTLA, Dr. G SUMALATHA, S ARUNA, NAGASIVA JYOTHI KOMPALLI, PALAMAKULA RAMESH BABU, SUNEETHA BANDEELA
Abstract:	Social media has become a vital tool for shaping public opinion, political debate, and consumer choice- making trend predictions and influencer analysis have never been more critical for businesses, policymakers and researchers. This paper introduces a new methodology of combining sentiment analysis, machine learning and network analysis to predict social media trends and identify cultural influencers. We acquire information from Twitter, Instagram, and Facebook using sophisticated NLP and machine learning models such as LSTM for Sentiment Classification, and ARIMA for Trend Prediction. A mixture of sentiment, user engagement and network centrality for the influential user is introduced. We have presented the results, showing the superiority of our hybrid approach against the traditional ones, with the MAE of trend forecasting for 0.145 and the influencer recommendation F1-score of 86.2%. The results demonstrate the capability of integrating sentiment analysis and influencer detection to enhance trend prediction and influence analysis. The methodology and its implementation provide useful tools for Real-time Social Media Analysis. They have applications in marketing, politics and social science by giving stakeholders control over social media dynamics.
Keywords:	Social Media, Trend Forecasting, Sentiment Analysis, Influencer Identification, Machine Learning, Network Analysis
Source:	Journal of Theoretical and Applied Information Technology 15^th August 2025 -- Vol. 103. No. 15-- 2025
Full Text

Title:	A LIGHTWEIGHT U-NET WITH SEPARABLE CONVOLUTIONS FOR EFFICIENT LUNG SEGMENTATION IN REAL-TIME MEDICAL IMAGING
Author:	SALEH ALGHAMDI, MOHAMMAD SHORFUZZAMAN
Abstract:	Accurate and efficient segmentation of the lung regions is indispensable for detecting and managing pulmonary diseases, as it allows clinicians to identify abnormalities and plan effective intervention strategies. However, the high computational demands of many existing segmentation models pose a significant challenge, particularly for deployment in resource-constrained environments such as mobile, edge platforms, and point-of-care devices. Lung segmentation is further challenged by wide anatomical variability and imaging artifacts, which existing models often struggle to handle without access to large-scale hardware. This study addresses this limitation by introducing a lightweight U-Net architecture that integrates depthwise separable convolutions to reduce computational complexity while preserving segmentation accuracy. By replacing standard convolutional layers, the model achieves faster inference and significantly lower parameter counts, making it well-suited for IT applications in embedded systems and clinical informatics. The model was evaluated on the publicly available Pulmonary Chest X-Ray Defect Detection dataset from Kaggle, demonstrating its effectiveness in segmenting lung regions. The performance evaluation shows that our model delivers outstanding results, attaining a Dice score of 91.92%, a Jaccard index of 82.75%, precision of 92.64%, recall of 90.31%, and accuracy of 97.12% on the test dataset. These results highlight that the lightweight U-Net achieves state-of-the-art segmentation accuracy with significantly reduced computational overhead, making it ideal IT solution for real-time use in clinical workflows and deployment on limited-resource devices.
Keywords:	Lung Segmentation, Depthwise Separable Convolutions, Lightweight U-Net, Chest X-ray Analysis, Real-Time Inference
Source:	Journal of Theoretical and Applied Information Technology 15^th August 2025 -- Vol. 103. No. 15-- 2025
Full Text

Title:	HYBRID MODEL FOR ASSESSING THE USABILITY OF A UNIVERSITY E-LEARNING PLATFORM: A CASE STUDY OF THE I-UH2C SYSTEM
Author:	MINA NKAILI, MAJIDA LAAZIRI, MOHAMED AZOUAZI
Abstract:	Promoting distance learning is essential to support students fresh from high school in their transition to full autonomy, but this transition must be implemented in practice. Gradually integrating distance learning into their curriculum is an effective approach to developing their self-learning skills. This method enables them to learn how to manage their time and assess themselves, thereby strengthening their autonomy and their ability to adapt to a variety of learning situations. This research aims to assess the usability and measurement of success and quality of learning platforms by developing a new evaluation approach based on two models and using a set of criteria developed from a quantitative and qualitative research methodology that was used at the Hassan II University in Casablanca and applied to the i-UH2C e-learning platform. The results obtained underline the importance of implementing an evaluation model for e-learning platforms to help the university achieve these objectives and improve the performance of such distance learning platform to meet the needs of learners. These results also encourage a better understanding of the positive aspects of this approach, such as active engagement in distance learning, ease of use of the platforms, interactivity and satisfaction endorsed by the majority of students, as well as areas that could be improved, including infrastructure, socio-economic disparities, and the learner’s socio-cultural environment.
Keywords:	Distance learning, platform, learning, evaluation, usability, quality, i-UH2C.
Source:	Journal of Theoretical and Applied Information Technology 15^th August 2025 -- Vol. 103. No. 15-- 2025
Full Text