		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
November 2024 | Vol. 102 No.22

Title:	THE EFFECT OF APPLIED ZAKAT ON WEALTH DISTRIBUTION USING AGENT BASED MODELING
Author:	HADDAD ABDERRAHIM , EL MOSAID FADMA
Abstract:	Zakat is the third pillar of Islam, therefore a mandatory alms and one of the most important tools of Islamic finance that ensures wealth redistribution and is used to reduce poverty in Muslim societies. Infact, it is intended to provide for the immediate needs of the eight categories of eligible recipients as they are recognized in the Holy Quran in Surat Tawba verse 60. The capitalistic system automatically generates a serious disparity between the rich and the poor then try to reduce this gap indirectly through taxation whereas the Zakat directly reaches the neediest and diminished category. In fact, the goal of the Zakat is to transform the poor into Zakat givers by pulling them out of poverty. This will be examined through simulation with agent based modeling commonly known as ABM and will be applied using simulation tools NETLOGO Version 6.4p by introducing the Islamic concept that is Zakat to the model of Simple Economy previously developed by Wilensky and used by [1] and [2] to study the Impact of Zakat on a Social Wealth Distribution. Many cases are simulated with and without Zakat and by using the indicators like Gini index, Lorenz curve, we are able to conclude that a wealth distribution is more equal and the gap between the rich and the poor is less significant compared to when Zakat is not applied.
Keywords:	Zakat, Agent Based, Wealth Distribution, Netlogo, Islamic Finance
Source:	Journal of Theoretical and Applied Information Technology 30^th November 2024 -- Vol. 102. No. 22-- 2024
Full Text

Title:	DEEP CNN BASED EMPIRICAL INVESTIGATIONS TO SKIN GRAZE UNCOVERING AND CATALOGUING USING HYBRID FEATURES SELECTION
Author:	DR. SURESH BABU CHANDOLU, DR.P. VENU MADHAV, DR. KANNEBOINA ASHOK, ANJANEYULU NAIK R, VENKATA NARAYANA T, G N.SOWJANYA, KURRA UPENDRA CHOWDARY
Abstract:	The proper disease diagnosis is one of the most important steps in medical treatment. In terms of diagnosis, one of the most unstable and difficult professions is dermatology. In order to make a correct diagnosis, dermatologists regularly need more patients because skin lesions, a deadly disease, can affect people of all ages. In order such as intelligent systems to diagnose skin cancer early and more accurately, skin lesion detection and categorization are essential. The term "multiclass skin lesions" refers to a group of subtypes of skin lesions, including basal cell carcinoma (BCC), melanocytic nevus (NV), melanoma (MEL), actinic keratosis (AK), benign keratosis lesion (BKL), squamous cell carcinoma (SCC), dermatofibroma (DF), and vascular lesion (VASC). The multi-class classifications are still a difficult task due to the wide range of skin lesions and their high similarities. It requires a significant amount of time, and expense to manually identify various skin lesions from dermoscopy images. Therefore, it is crucial to develop automated diagnostics techniques that can more accurately classify skin lesions of multiple classes. Hence this study presents Deep Convolutional Neural Network (DCNN)-based hybrid feature selection is used for multiclass skin lesion detection and classification. The sensitivity, accuracy, and specificity of the architecture that is being provided are used to evaluate its performance.
Keywords:	Dermatology, Skin cancer, Skin lesion, Hybrid features selection and DCNN
Source:	Journal of Theoretical and Applied Information Technology 30^th November 2024 -- Vol. 102. No. 22-- 2024
Full Text

Title:	EMOTION ANALYSIS OF TEXTUAL CONTENTS USING NATURAL LANGUAGE PROCESSING AND TEXT MINING
Author:	SUBHASH CHANDRA GUPTA, NOOPUR GOEL
Abstract:	Background: As the internet based e-commerce applications are expanded, a tsunami of comments are arised in the digital world. Most of the comments are textual in nature, and a reflection of emotions made by the people from their day-to-day activities performed on these applications. Analysis of these comments is not an easy task, a highly technological model based on machine learning classifiers is required to perform this task. Methods: Model's Performance is based on the classifiers, the higher the count of correct predictions made by the classifier, the better performance the model shows. In the proposed NLP based model, seven classifiers used in the model are KNN, Logistic regression, SVM, multinomial Naive Bayes, decision tree, random forest and Gradient Boosting. Input dataset is an example of multiclass, during the preprocessing, noise was removed from the dataset and converted into a binary class dataset. Model uses the term frequency-inverse document frequency (TF-IDF) to vectorize it for feature extraction before applying classifiers to produce its results. Result and Analysis: Finally a comparative study has been made to analyze the obtained results of all classifiers used in the model. From the analysis, it is found that the SVM classifier is the best scorer for the model and it scores accuracy, F1 score, precision, recall and AUC 96.42%, 96.38%, 96.42%, 96.15% and 96.35 % respectively.
Keywords:	Sentiment Analysis, Imbalanced dataset, NLP, Text Mining, Text summarization.
Source:	Journal of Theoretical and Applied Information Technology 30^th November 2024 -- Vol. 102. No. 22-- 2024
Full Text

Title:	OPTIMIZED ANN HYPERPARAMETERS TO IDENTIFY MALICIOUS TRAFFIC IN NETWORKS
Author:	HIND KHOULIMI, OTHMAN BENAMMAR
Abstract:	Intrusion detection systems have been a critical research area for over three decades. With the growth of internet traffic, the number of attacks that violate the confidentiality, integrity, and authenticity of important data has increased significantly. The advent of Artificial Intelligence (AI) especially Deep Learning (DL) creates models automatically to detect malicious traffic without human intervention. In this context, we have proposed an Intelligent Security Management System (ISMS) based on detection, analysis, and action engines. To guarantee efficiency and accuracy, we have used Artificial Neural Network (ANN) as a classification model, and to achieve better accuracy, different optimization algorithms are applied to select the best hyperparameters (weights and biases) for our model. In this present paper, we have based our study on KDD CuP 99, NSL KDD, and UNSW-NB 15 dataset to evaluate the best combination of ANN and optimization algorithms, using metrics: accuracy, loss, training time, precision, recall, and F1-score.
Keywords:	Artificial Neural Network, Harris Hawks Optimization, Particle Swarm Optimization, Spider Monkey Optimization; Cat Swarm Optimization.
Source:	Journal of Theoretical and Applied Information Technology 30^th November 2024 -- Vol. 102. No. 22-- 2024
Full Text

Title:	A SYSTEM FOR ANALYSING CALL DROP DYNAMICS IN THE TELECOM INDUSTRY USING MACHINE LEARNING AND FEATURE SELECTION
Author:	CHEEPU BALAKRISHNA, DR CINDHE RAMESH , S MEGHANA, DR. C. DASTAGIRAIAH
Abstract:	Quality of service in the telecom industry plays a vital role in the growth and economy of the country. In India, several telecom operators provide services, and there is a regulatory authority known as the Telecom Regulatory Authority of India (TRAI). In the telecom domain, call dropping is a problem that deteriorates the performance of the telecom industry in rendering services. It also causes inconvenience and waste of time for users, as well as reducing the level of user satisfaction. There is a need for a technology-driven analysis with the help of Artificial Intelligence (AI) to analyze call drop dynamics toward making well-informed decisions. The existing research revealed that Machine Learning (ML) helps analyze call drop dynamics. However, there is a need for a framework with machine learning techniques and optimizations to improve performance in analyzing call drop dynamics in the telecom industry. In this paper, we proposed an ML framework for automatic analysis of all drops in the telecom industry across all operators. The framework also supports optimizations like future engineering and dimensionality reduction to improve the performance of machine learning models. We proposed an algorithm called Learning based Call Drop Analytics (LbCDA) which exploits feature selection and training multiple classifiers towards call drop analytics. With benchmark dataset variants of the telecom industry, our empirical study has revealed that our framework's Random Forest (RF) model outperforms other models with the highest accuracy of 87.40%.
Keywords:	Telecom Industry, Call Drop Analysis, Random Forest, Machine Learning, Artificial Intelligence
Source:	Journal of Theoretical and Applied Information Technology 30^th November 2024 -- Vol. 102. No. 22-- 2024
Full Text

Title:	AN EXTENSIVE REVIEW OF SECURITY ISSUES AND CHALLENGES IN FOG COMPUTING ENVIRONMENT
Author:	W. ALGESHARI, M. SHER. RAMAZAN, F. ALOTAIBI, K. ALYOUBI
Abstract:	As a crucial addition to cloud computing, fog computing provides localized processing resources and services near end devices at the network edge. The drawbacks of traditional cloud computing, like excessive latency, immobility, and poor location awareness, are addressed by this design, which is especially advantageous for the Internet of Things (IoT). Despite its advantages, fog computing poses serious security risks because of its decentralized architecture, close proximity to end users, and confined processing power. These security issues become more serious as IoT device counts rise, increasing the possibility of cyberattacks. This review article offers an extensive overview of the security issues and challenges unique to fog computing systems. It outlines various attack vectors, examines the design, traits, and vulnerabilities of fog nodes, and addresses mitigating these risks with countermeasures. This paper attempts to identify important security issues in fog computing and suggest future strategies for resolving these issues to ensure safe and reliable fog-IoT ecosystems by assessing previous research.
Keywords:	IOT, Cloud Computing, FOG Computing, FOG Attacks
Source:	Journal of Theoretical and Applied Information Technology 30^th November 2024 -- Vol. 102. No. 22-- 2024
Full Text

Title:	APPLICATION OF THE SIMPLE ADDITIVE WEIGHTING METHOD IN THE PERFORMANCE ASSESSMENT OF ENERGY COMPANIES BASED ON ROA AND DAR
Author:	ENI DUWITA SIGALINGGING, ERLINA, ISKANDAR MUDA, DIO AGUNG HERUBAWA
Abstract:	Performance evaluation of energy companies in Indonesia is very important to support efficient and sustainable energy sector management. One way to evaluate company performance is to use the Simple Additive Weighting (SAW) method. This method combines several criteria in one comprehensive assessment system. This study is to test the use of the SAW method in assessing the performance of energy companies in Indonesia. The focus is on two main indicators: Return on Assets (ROA) and Debt to Asset Ratio (DAR). Data for this study were taken from energy companies listed on the Indonesia Stock Exchange, which were obtained from the company's financial statements. The SAW method is used by giving weights to the criteria, normalizing the data, and calculating the ranking based on the normalized decision matrix. The research is expected to help a more efficient and accurate decision support system in assessing the performance of energy companies. In addition, it is also expected to support government policies in managing the energy sector sustainably. It is hoped that using the SAW method can make the evaluation process faster, more efficient, and more accurate than using the manual method that is usually used today.
Keywords:	Debt to Asset Ratio, Performance assessment, Return on Assets, Simple Additive Weighting.
Source:	Journal of Theoretical and Applied Information Technology 30^th November 2024 -- Vol. 102. No. 22-- 2024
Full Text

Title:	OPTIMIZING CROP YIELD PREDICTION CROP YIELD PREDICTION: A HYBRID APPROACH INTEGRATING CNN AND LSTM NETWORKS
Author:	HUI HUI WANG, YIN CHAI WANG, BUI LIN WEE, JANE KHOO YAN, FARASHAZILLAH BINTI YAHYA
Abstract:	Artificial Intelligence (AI) has proven successful in revolutionizing the agricultural sector, facilitating advancements in prediction, decision-making, and the monitoring and analysis of crops and soil. In this study, a hybrid model is introduced with the capability to predict crop yield. The proposed learning model combines the strengths of Convolutional Neural Network (CNN) with Recurrent Neural Network (RNN) models. CNN, recognized for its superior performance in feature extraction, is selected for its characteristic of considering a smaller number of parameters in the network, thereby reducing the risk of overfitting. Simultaneously, RNN serves as the prediction model, capitalizing on its inherent learning nature, feedback network, and ability to encode temporal sequence information. Addressing the short-term memory behaviour of RNN, the network is enhanced with LSTM cells, enabling effective long-term memory tasks. LSTM introduces memory blocks to resolve the exploding and vanishing gradient problem, differentiating itself from conventional RNN units. The best environment parameters have been identified by using the correlation where it shows the parameter that have the most significant relation with the crop production. The A Hybrid Approach Integrating CNN and LSTM Networks has achieved 74% accuracy in crop yield prediction.
Keywords:	Agriculture, Convolutional Neural Network, Crop Yield Prediction, Machine Learning, Recurrent Neural Network (RNN)
Source:	Journal of Theoretical and Applied Information Technology 30^th November 2024 -- Vol. 102. No. 22-- 2024
Full Text

Title:	A SCALABLE AND SECURE BLOCKCHAIN-BASED HEALTHCARE SYSTEM: OPTIMIZING PERFORMANCE, SECURITY, AND PRIVACY WITH ADAPTIVE TECHNOLOGIES
Author:	P. VINAYASREE, A. MALLIKARJUNA REDDY
Abstract:	This Paper Presents A Scalable, Secure Blockchain-Based Healthcare System Architecture That Efficiently Manages Extensive Patient Data While Ensuring High Security. Adaptive Partitioned Filters (APFS) and Compact Patricia Tries (CPTS) Enable Efficient Data Access and Management, While Sharded Byzantine Optimized Consensus (SBOC) and Go's Concurrency Model Facilitate Parallel Transaction Processing. Security is Provided Through Bloom Filters, Patricia Tries Extended By Merkle Trees, and an Immutable Blockchain Ledger Protected by Practical Byzantine Fault Tolerance (PBFT). Verifiable Random Functions (VRF) Secure Participant Selection for Consensus, and Zero-Knowledge Proofs (ZK-SNARKS) Verify Transactions without Revealing Sensitive Information, Aligning With Healthcare Regulations. Chacha20 Encrypts Sensitive Data, and Role-Based Access Control (RBAC) Governs Access Rights. This Architecture offers a Comprehensive Solution for Scalable, Efficient, and Secure Healthcare Data Management in Blockchain Environments.
Keywords:	Blockchain, Adaptive Partitioned Filters (APFs), Compact Patricia Tries (CPTs),Sharded Byzantine Optimized Consensus (SBOC), Zero-Knowledge Proofs (zk-SNARKs), ChaCha20 encryption, Role-Based Access Control (RBAC), Verifiable Random Functions (VRF).
Source:	Journal of Theoretical and Applied Information Technology 30^th November 2024 -- Vol. 102. No. 22-- 2024
Full Text

Title:	EVOLVING INTERESTS AND PREFERENCES USING ARIMA AND STL IN SOCIAL MEDIA SEARCH BEHAVIORS
Author:	B. RAMESH BABU, S. RAMAKRISHNA
Abstract:	This study investigates the nuances of user search behavior on social media platforms by employing sophisticated time series algorithms: the Autoregressive Integrated Moving Average (ARIMA) and Seasonal Decomposition of Time Series (STL). In an era where social media constitutes a vital conduit for information dissemination and interaction, understanding the patterns of user searches and engagement becomes crucial. This research aims to dissect the multifaceted nature of search behavior, focusing on trends, user engagement, and the popularity of content. Through the integration of ARIMA and STL models, we meticulously analyze the temporal aspects of these behaviors, unveiling the progression of user interests and preferences. The research methodology encompasses a comprehensive approach to data collection and preprocessing, followed by the application of these advanced algorithms to model and scrutinize search patterns effectively. Additionally, the study examines the influence of external factors on search behavior and evaluates algorithms for content recommendation, aiming to optimize content delivery and bolster user engagement on social media platforms. Notably, the proposed method of combining ARIMA with STL has demonstrated a significant improvement in predictive accuracy, surpassing traditional models. Specifically, the ARIMA+STL model showed remarkable enhancements, reducing the Mean Absolute Error (MAE) by approximately 67% and the Root Mean Squared Error (RMSE) by nearly 81% compared to baseline models. Moreover, the Akaike Information Criterion (AIC) was notably lower, indicating a superior model fit with optimized complexity. These findings underscore the effectiveness of our approach in capturing the intricate dynamics of social media search behavior, offering valuable insights for developing more refined digital engagement strategies.
Keywords:	Social Media Analytics, ARIMA Model, STL Decomposition, User Engagement, Search Behavior Patterns
Source:	Journal of Theoretical and Applied Information Technology 30^th November 2024 -- Vol. 102. No. 22-- 2024
Full Text

Title:	THE DETERMINANTS OF SOFTWARE DEVELOPERS’ INTENTION TO ADOPT CHAT GPT AS PROGRAMMING INFORMATION MEDIA
Author:	IFVO DEKY WIRAWAN, TANTY OKTAVIA
Abstract:	In the digital era, technology is developing rapidly. One of them in the field of artificial intelligence is chatbots. A chatbot is a computer program designed to stimulate conversation or interactive communication with (human) users via text, voice, or visuals. One of the chatbots that is currently popular is ChatGPT. ChatGPT training models interact in a conversational manner with a dialogue format that allows ChatGPT to answer follow-up questions, admit mistakes, violate false premises, and reject inappropriate requests. Software Developers are one of the users who can take advantage of this ChatGPT technology tool. There are several problems that software developers often experience in the coding process, including code or algorithms that are too complicated, making the code difficult for other people to understand and problems in the code that are sometimes difficult to overcome. The purpose of this research is to analyze what factors influence software developers' intentions and adoption of using ChatGPT as an information media in programming. This research model is UTAUT. This study involved surveying 399 participants through the distribution of questionnaires. The data collected was then processed and analyzed utilizing the PLS-SEM method. The findings revealed that factors such as Performance Expectancy, Effort Expectancy, Trust, Perceived Risk, and Experience significantly influenced Behavioral Intention. Furthermore, Behavioral Intention was identified as having a significant impact on the Intention to Adopt. Meanwhile, Social Influence and Facilitating Conditions do not have a significant effect on Behavioral Intention.
Keywords:	Artificial Intelligence, Chatbot, ChatGPT, SMART PLS, UTAUT
Source:	Journal of Theoretical and Applied Information Technology 30^th November 2024 -- Vol. 102. No. 22-- 2024
Full Text

Title:	THE ROLE OF CITIZEN ENGAGEMENT IN DEMOCRATIC GOVERNANCE ENHANCEMENT THROUGH E-GOVERNANCE: A CASE STUDY OF LUSAKA CITY COUNCIL, ZAMBIA
Author:	LAMECK NSAMA, CHIMEKO KENNY WEBSTER, BIBHUTI BHUSAN DASH, NGULA WALUBITA, UTPAL CHANDRA DE, SUDHIR KUMAR BEHERA, SATYENDR SINGH, SUDHANSU SHEKHAR PATRA
Abstract:	The study investigates how e-governance could be used to foster the engagement of citizens in the process of democratic local governance with an emphasis on the Lusaka City Council. Democratic governance involves principles of openness, responsibility, and involvement. The introduction of e-governance brings new opportunities for improving indirect communication between the government and the citizens. This paper seeks to assess the extent of the impact of executed e-governance projects in Lusaka City Council on democratic politics and results. The data source is also diverse based on the view that both qualitative and quantitative data will be used in the study. Interviews with local inhabitants and councilors are conducted to educate the authors about their experiences with and revelations about e-governance. Furthermore, a content audit and an analysis of the documents and e-governance applications existing in LCC is carried out to examine the ways of its accessibility and utility as well as its effects. From research carried out on the topic, it can be deduced that e-governance has enhanced citizens’ engagement in local governance. Availability and use of information through digital platforms have boosted politically responsive governance output in that it has raised the standards of accountability. Also, other forms of communication that have come with the use of the Internet include discussion forums and avenues of giving feedback online has allowed citizens to give more opinions on any issue or decision-making process. The study also outlines some barriers including; digital literacy and digital infrastructure as factors that influence the success of e-governance programs in Zambia. Thus, the research indicates that while e-governance has the capability to enhance democratic accountability at the grassroots, its effectiveness remains contingent on the existing challenges’ eradication and enabling everyone to engage. The case of the Lusaka City Council teaches other municipalities noble lessons on the prospects of e-governance regarding citizens’ engagement and the fortunes of democracy.
Keywords:	Lusaka, Citizen Engagement, E-governance, Public Service, Democratic e-governance, Participatory governance
Source:	Journal of Theoretical and Applied Information Technology 30^th November 2024 -- Vol. 102. No. 22-- 2024
Full Text

Title:	INSIGHTS INTO METADATA COMPONENTS: A SYSTEMATIC REVIEW OF ENTERPRISE DATA CATALOGS
Author:	ERIC WIJAYA, TANTY OKTAVIA
Abstract:	In the contemporary data-centric business environment, proficient metadata and data documentation processes are essential for organizations aiming to maximize the value of their data assets. This data-driven architecture, when coupled with Business Intelligence (BI) tools, promotes data democratization, allowing stakeholders throughout the business to utilize it. This research examines the exploration of metadata elements on the development of BI systems. A thorough literature research and a preliminary analysis are done to comprehend the landscape of metadata classification. This study delves into existing research on metadata and data catalogue management in enterprises, utilizing a systematic literature review (SLR) to identify specific metadata components. The SLR results not only describe the functions of each metadata component, but also provide practical guidance on how to adopt them, making them helpful insights for firms wishing to enhance their data management. Along with the bibliometric study investigates trends and partnerships among metadata, providing further information on efficient metadata implementation. These findings have implications for firms looking to improve their data processes and achieve a competitive advantage by providing new insights into management tactics, opening the way for future research in metadata and data catalogue systems.
Keywords:	Metadata, Systematic Literature Review, Bibliometric, Catalogue System
Source:	Journal of Theoretical and Applied Information Technology 30^th November 2024 -- Vol. 102. No. 22-- 2024
Full Text

Title:	AN INTELLIGENT MODEL FOR INTEGRATING ENTERPRISE DATA WAREHOUSE LAYERS TO MANAGE SCHEMA EVOLUTION
Author:	WADEA GEORGE , MOHAMMED MARIE , AHMED YAKOUB
Abstract:	In contemporary organizations, data warehousing centralizes and manages data from diverse sources, such as relational databases and semi-structured systems like Temenos core banking (T24) XML systems, facilitating data analytics and informed decision-making. However, the dynamic nature of data and evolving source schemas pose challenges in adapting data warehouses efficiently, leading to interruptions in data loading processes. This research proposes an intelligent model to enhance data integration in data warehouses, aiming to automate or semi-automate the development cycle and integrate various layers, including business analysis, data modeling, ETL (Extract, Transform, Load), and data quality enhancement through rejection handling. Leveraging metadata, data dictionaries, data mining techniques, and Data Vault modeling, the model aims to reduce development time and costs. The proposed model offers an efficient solution to adapt to changes, significantly reducing adaptation costs compared to prior approaches. By seamlessly integrating all layers of the data warehouse (DWH), it streamlines development cycles through automation or semi-automation, introducing additional features to expedite the process. This research demonstrates the model's superiority through three implemented experiments, showing significant time savings and cost reductions (75% decrease compared to manual processes). It successfully integrates development layers, semi-automates the development process, filters rejected data, and provides a clear vision of schema storage during new changes deployment.
Keywords:	ETL; DWH; Data modeling; Schema evolution; Information Retrieval; Metadata.
Source:	Journal of Theoretical and Applied Information Technology 30^th November 2024 -- Vol. 102. No. 22-- 2024
Full Text

Title:	FREQUENCY STABILIZATION WITH SOLAR PV INTEGRATION IN TWO-AREA INTERCONNECTED MICRO GRID SYSTEM
Author:	ENDOORI RANI, Dr. K. NAGA SUJATHA
Abstract:	Microgrids (MGs) are becoming more reliant on Renewable Energy Sources (RESs) to meet consumer demand but the fluctuating output of RESs, combined with the unpredictable behaviour of loads can contribute to frequency instability. The incorporation of additional renewable energy sources enhances the system's inertia, contributing to frequency stabilization. This work examines the frequency regulation (FR) in a microgrid system with two-area control, incorporating a variety of energy sources such as electric vehicles (EVs), fuel cells, wind turbines, energy storage systems, conventional generators, and solar PV systems. A solar photo voltaic (PV) system, equipped with an inverter was integrated into the existing microgrid for evaluation of microgrid’s performance and analysed its operation with and without the PV system, specifically examining deviations in frequency and power. Classical controllers, including PI, PID, and I, are implemented in Area-1 and Area-2 of the microgrid to enhance frequency and power stability. Magnitude of frequency and power fluctuations were tabulated to compare the performance in Area-1 and Area-2 of Micro Grid with photo voltaic (PV) and without PV systems.
Keywords:	Solar PV System, Frequency Regulation, Storage System, Distributed Energy Systems, Renewable Energy Sources, Single Area System, Two-Area Interconnected Syste
Source:	Journal of Theoretical and Applied Information Technology 30^th November 2024 -- Vol. 102. No. 22-- 2024
Full Text

Title:	LSTM NEURAL NETWORKS WITH BAYESIAN OPTIMIZATION FOR REORDER SIMULATION IN RETAIL INVENTORY MANAGEMENT
Author:	INTAN RAHMATILLAH, IMAN SUDIRMAN, ANTON MULYONO AZIS, IVAN DIRYANA SUDIRMAN
Abstract:	This research tries to improve the accuracy of demand forecasting in grocery store with optimization of LSTM neural networks by using Bayesian Optimization. Research Design Data and Methodology: This study aim to forecast the demand by using Brand AB instant noodles daily sales data from a grocery store in one year period. Well-tuned by Bayesian optimization, LSTM model can encode intricate temporal structure and long-term dependency in sales sequences. Result: The results show that the optimized LSTM model has a low value of mean squared error (MSE) 0.0056, which means good predictability, a simple reorder simulation incorporates the enhanced forecast model to refine optimal inventory management decisions. Through this simulation, critical reorder points will be established to ensure that the product is always available and there is no risk of stock outs or overstocked items. Conclusion: The results in this study demonstrate a great potential of sophisticated neural networks and the Bayesian optimization approach to achieve better inventory management within retail industry.
Keywords:	LSTM Neural Networks, Bayesian Optimization, Inventory Management, Reorder Simulation
Source:	Journal of Theoretical and Applied Information Technology 30^th November 2024 -- Vol. 102. No. 22-- 2024
Full Text

Title:	OPTIMIZED EFFICIENTNET WITH GENETIC EXPRESS PROCESSING FOR ACCURATE LUNG DISEASE CLASSIFICATION
Author:	YELLEPEDDI SAMBA SIVA KRISHNA ASSISH AND KUPPUSAMY P
Abstract:	The ongoing COVID-19 pandemic underscores the urgency for rapid and precise diagnostic tools. This study presents an innovative approach for classifying lung diseases COVID-19, asthma, and pneumothorax using Computed Tomography (CT) lung images. The EfficientNet B4 is proposed to classify lung diseases accurately using compound scaling features including depth, width, and resolution. By integrating the EfficientNet model with a Genetic Express Processing Algorithm (GEP) for hyperparameter tuning, the proposed method focuses on optimizing dropout, learning rate, and batch size. Fine-tuning the EfficientNet B4 model through compound scaling and hyperparameter optimization led to a classification accuracy of 96.5%. Visualizing lung-infected regions using Class Activation Maps (CAMs) provides insights into classification decisions. This research work incorporates Generative Adversarial Networks (GANs) to generate synthetic images that enhances data diversity and model generalization. This method combines Deep Learning (DL) models with Genetic Algorithms (GA) and GANs, demonstrating substantial improvements in disease detection accuracy. The proposed approach offers medical professionals efficient diagnostic tools for early and reliable disease diagnosis. Code can be available at https://github.com/YellepeddiSambaSivaKrishnaAssish/Optimized-EfficientNet-using-GEP-for-Lung-diseases.git.
Keywords:	Computed Tomography, COVID-19, Genetic Express Processing, Artificial Intelligence, Optimization, Deep Learning
Source:	Journal of Theoretical and Applied Information Technology 30^th November 2024 -- Vol. 102. No. 22-- 2024
Full Text

Title:	INTERACTIVE LEARNING PLATFORMS TO STEAM (I-LPS) GAMIFICATION FOSTERING COMPUTATIONAL INNOVATORS AND CREATIVE THINKING IN TEACHER EDUCATION
Author:	PONGSATON PALEE, JITTIMA PANYAPISIT, ADIREK YAOWONG, THIPWIMOL WANGKAEWHIRAN, PANITA WANNAPIROON
Abstract:	The purpose was to develop and study aims to analyze, synthesize, design and develop interactive learning platforms with STEAM (I-LPS) gamification fostering computational innovators and creative thinking in teacher education. The research instruments included a manual for I-LPS, active learning lesson plans, an active citizenship competencies test, and a satisfaction questionnaire. Statistics for data analysis were percentage, mean, standard deviation, and dependent sample t-test. The research is an application of the concept of Research and Development (R&D) and defines the framework for conducting research into 5 phases: Phase 1 (R1) Study and synthesis of the conceptual framework for the development of interactive learning innovative media with a simplified active learning model. Phase 2 (D1) Conceptual Framework for Developing Interactive Learning Materials with Active Learning Model Phase 3 (R2) Evaluation of the Conceptual Framework of the Gamified Digital Learning Ecosystem by Asking for Expert Opinions Phase 4 (D2) Creating Interactive Learning Materials with Gamified Active Learning Model to Promote Innovative Skills Phase 5 (R3) Evaluation of Achievement and Satisfaction. The overall student satisfaction with the interactive learning materials remains was high (mean = 4.29, S.D. = 0.65) and indicating that the results of the measurement before studying and the learning achievement after studying with the normal teaching method were statistically significantly different at the .01 level.
Keywords:	Interactive Learning Platform, STEAM, Computational Innovators, Creative thinking
Source:	Journal of Theoretical and Applied Information Technology 30^th November 2024 -- Vol. 102. No. 22-- 2024
Full Text

Title:	RECURRENT NEURAL NETWORK-BASED BRAIN TUMOR CLASSIFICATION USING 3D MAGNETIC RESONANCE IMAGING
Author:	Dr. S. KOTHER MOHIDEEN, K. A. MOHAMED RIYAZUDEEN
Abstract:	Brain tumor identification and segmentation using magnetic resonance imaging (MRI) represent difficult but critical tasks for a variety of applications in the area of medical analysis, including cancer detection and treatment. Brain tumor implies the collection of aberrant cells in certain brain tissues. The brain tumor may be malignant or cancer-free. Glioma, Meningioma are the most frequent kinds of brain tumors. Early identification of tumor cells plays an important role in patient therapy and recovery. Brain tumor diagnosis is typically subject to a highly complex and time-consuming procedure. MRI scans of different patients may be utilized to identify cancers at various stages. There are many kinds of functional extraction and classification techniques used to identify brain tumors from 3D MRI brain images. Specifically, in this article, we offer a preprocessing method that works just on a small portion of the image rather than the whole image to produce a flexible and efficient brain tumor segmentation system. This technique reduces the amount of computation time required and eliminates the overfitting issues that may occur in a deep-learning model. In the second phase, we suggest a simple and effective Recurrent Neural Network (RNN) while we are interacting with a smaller portion of brain images in each slice. The Recurrent neural network algorithm exploits combine local and global characteristics and it does so in two distinct ways. Additionally, it enhances the accuracy of brain tumor segmentation when compared to current state-of-the-art models. The image classification method for the RNN enables the early detection of the tumor with great accuracy. We presented a recurring neural network design for tumor cell identification, which is about 99.92% accurate. A recurring neural network (RNN) is a kind of artificial neural network where connections between nodes create a graph in a time sequence.
Keywords:	Magnetic Resonance Image, Malignant, Glioma, Meningioma, Recurrent Neural Network
Source:	Journal of Theoretical and Applied Information Technology 30^th November 2024 -- Vol. 102. No. 22-- 2024
Full Text

Title:	EXPLORING MACHINE LEARNING METHODS FOR INTRUSION DETECTION SYSTEM: A DEEP DIVE INTO TECHNIQUES, DATASETS, AND PERSISTENT CHALLENGES
Author:	SHAIK JOHNY BASHA, D. VEERAIAH, SUMALATHA LINGAMGUNTA
Abstract:	Intrusion Detection Systems (IDS) play a crucial role in safeguarding modern digital infrastructures by identifying potential threats and anomalies in real time. As cyberattacks become more sophisticated, leveraging Machine Learning (ML) techniques in IDS has emerged as a promising approach to enhance detection accuracy, adaptability, and resilience. This paper provides an in-depth exploration of various ML methods applied to IDS, categorizing techniques such as supervised, unsupervised, and reinforcement learning. Additionally, it delves into the most used datasets for training and evaluating IDS models, highlighting their characteristics, advantages, and limitations. Furthermore, the paper addresses the persistent challenges in deploying ML-driven IDS, including issues related to data imbalance, real-time performance, adversarial attacks, and model generalization. Through a comprehensive analysis of current research and future directions, this survey aims to offer insights into the evolving landscape of ML-based IDS, paving the way for more robust and scalable solutions in the face of ever-evolving cyber threats.
Keywords:	IDS Dataset, Cyber security, Machine Learning, Intrusion Detection System (IDS), Network Security
Source:	Journal of Theoretical and Applied Information Technology 30^th November 2024 -- Vol. 102. No. 22-- 2024
Full Text

Title:	MULTI-CLASS PLANT DISEASE CLASSIFICATION AND STAGE-WISE SEVERITY PREDICTION BASED ON MULI-CLASS LEAF LABELLING
Author:	D.LITA PANSY, M.MURALI
Abstract:	Plant Disease Severity Prediction (PDSP) aids in increasing the yield of a plant. However, research has been done scarcely on the severity prediction of multi-class Plant Disease (PD). Hence, this paper proposes multi-class leaf labeling and Exponential Pareteo Fuzzy (EP-Fuzzy) based disease severity prediction. Primarily, input images with the complex background are taken and the background is removed in those images. The obtained background-removed images are utilized for increasing the dataset via augmentation and providing labeled data for the leaf labeling model utilizing augmented images. The unlabelled images are labelled in the Spearman-based Pseudo Labelling (S-PL) model by utilizing the augmented images. The labelled images and saliency-mapped images are fused to enhance the Segmentation Accuracy (SA) of Seam Carving-Region Split and Merge (SC-RSM). Afterward, the Depth-wise ResNet-50 (DRN) classified the class of the leaf diseases, with the segmented image and extracted feature. By utilizing the EP-Fuzzy model, the severity stage is predicted for the classified diseased leaf. The proposed technique’s performance is experimentally assessed, where the proposed one exhibited superior performance on accuracy, training time, overall prediction rate, Dice Similarity Coefficient (DSC), et cetera.
Keywords:	Plant Disease Severity Prediction (PDSP), Multi-class Leaf label, saliency maping, Spearman-based Pseudo labelling (S-PL), Depth-wise ResNet-50 (DRN).
Source:	Journal of Theoretical and Applied Information Technology 30^th November 2024 -- Vol. 102. No. 22-- 2024
Full Text

Title:	DATA-DRIVEN INSIGHTS INTO FISHING PATTERNS USING VMS AND MACHINE LARNING
Author:	KANDA MAHENDRA, TANTY OKTAVIA
Abstract:	The fishing industry plays a crucial role in global food security, yet continues to face significant monitoring and regulatory challenges. One of the most pressing issues is accurately tracking fishing vessel behavior, especially given the rising concerns of illegal, unreported, and unregulated (IUU) fishing that threatens marine ecosystem sustainability. Current monitoring systems often struggle to reliably distinguish between legitimate fishing operations and suspicious activities. To address this challenge, our study introduces an innovative approach combining Vessel Monitoring System (VMS) data with a Hidden Markov Model (HMM) to track and analyze vessel movements. This method focuses on vessel speed patterns to identify different fishing activities including hauling, traveling, and active fishing. To strengthen the accuracy of our analysis, we enhanced the HMM approach by incorporating three complementary machine learning techniques: Naive Bayes, Support Vector Machine (SVM), and Gradient Boosting Machine (GBM). This combined approach allows us to better understand and classify various fishing activities by examining speed patterns and movement transitions. Our results demonstrate significant improvements in detecting and classifying fishing activities, particularly in distinguishing between different phases of fishing operations and identifying unusual patterns that might indicate illegal activities. The study concludes that this integrated approach substantially improves our ability to monitor fishing activities, with notably higher accuracy rates in classification. These findings offer promising implications for fisheries management, providing a practical and effective way to monitor fishing activities and promote sustainable practices. Our framework offers a flexible and powerful tool for fisheries regulators, helping them better protect marine resources through improved surveillance and monitoring capabilities.
Keywords:	Vessel Monitoring System (VMS), Hidden Markov Models (HMM), Naive Bayes, Gradient Boosting Machine (GBM), Support Vector Machine (SVM), fishing trip behavior, machine learning, fisheries management, environmental factors, marine resources management.
Source:	Journal of Theoretical and Applied Information Technology 30^th November 2024 -- Vol. 102. No. 22-- 2024
Full Text

Title:	ENHANCING PATIENT MONITORING IN WIRELESS BODY AREA NETWORK THROUGH SMA-INTEGRATED CONVOLUTIONAL NEURAL NETWORK
Author:	MD. TASLIM AREFINN, MD ABUL KALAM AZAD
Abstract:	Sophisticated systems of surveillance that keep tabs on life’s essential functions are called health maintenance technologies. The goal of the endeavor was to plan and construct wirelessly bodily networks of sensors for real-time performance assessment. WBANs need to analyze massive volumes of data for the purpose of making practical judgments during emergencies. In order to overcome these problems, this study presents a deep learning structure for evaluating health consequences called the Slime Mould Algorithm (SMA). In the beginning, information from medical records for patients is gathered by the WBAN networks in order to produce specific measurements for the assessment. WBAN modules communicate with the destination node by sending information based on the collected indicators. In this scenario, the optimal cluster head is determined using the Fruit Fly technique. The combined Fruit fly Procedure’s results are then sent to the destination component, whereupon the Convolutional Neural Network, also known as the classifies the medical data in order to assess risk. In this instance, the CNN is trained using the recommended SMA. With scores of 94.604% and 0.145, along with 0.058 for accuracy, power, and productivity, correspondingly, the suggested SMA outperforms the other methods.
Keywords:	Patient Monitoring System, Convolutional Neural Network, Slime-Mould Approach, Wireless Body Area Network, Fruit-Fly
Source:	Journal of Theoretical and Applied Information Technology 30^th November 2024 -- Vol. 102. No. 22-- 2024
Full Text

Title:	ADVANCING CROP YIELD PREDICTION THROUGH MACHINE AND DEEP LEARNING FOR NEXT-GEN FARMING
Author:	UDAYA KUMAR ADDANKI , TEJASWI MADDINENI , VIJAY DHAWALE , M.L.M.PRASAD , DESIDI NARSIMHA REDDY , JEEVAN JALA
Abstract:	Agriculture has contributed to India's GDP, accounting for 15-18% of the economy. However, Indian agriculture faces persistent challenges threatening its long-term stability, including soil degradation, pest management issues, and fluctuating crop prices. These challenges create significant uncertainty in crop yields. To address this, we propose data-driven solutions using machine learning and deep learning models to improve the accuracy of crop yield predictions. Machine learning models, such as decision trees, random forests, gradient boosting, and ensemble techniques like XGBoost, along with deep learning models like convolutional neural networks (CNNs) and long short-term memory (LSTM) networks, provide reliable predictions and precise forecasting, enabling farmers to achieve more stable and optimized yields. Beyond economic benefits, accurate crop prediction also enhances food security and strengthens rural economies. By advancing precision in agricultural forecasting, these methods can help tackle longstanding agricultural issues, contributing to economic growth, increased profits, and a stable food supply for India. Embracing data-driven approaches is essential to addressing the evolving challenges of the nation's agricultural sector.
Keywords:	Deep learning, Machine learning, Agriculture, LSTM, CNN, Random Forest, Decision tree, Gradient Boosting, XGBoost
Source:	Journal of Theoretical and Applied Information Technology 30^th November 2024 -- Vol. 102. No. 22-- 2024
Full Text

Title:	CUSTOMER SEGMENTATION IN THE ONLINE RETAIL INDUSTRY USING BIG DATA ANALYTICS
Author:	RONALD S. CORDOVA
Abstract:	Nowadays, customer data is abundant due to the growth of the online retail industry. It allows effective customer segmentation using big data analytics. This study examines how big data analytics segments online customers. Using segmentation algorithms and data processing to optimise marketing, sales, and customer experiences. Different customer segments can be identified by traits and behaviours. This research examines customer segmentation theory in the online retail industry. Data collection and preprocessing methods are discussed to ensure data quality and segmentation relevance. This study will also demonstrate how customer segmentation strategies can be used to drive digital marketing and sales campaigns and improve customer experiences using big data analytics. Furthermore, this research will demonstrate big data analysis output using simulations. It will illustrate how big data analytics can segment customers. This will help online retailers tailor their marketing and sales to each cluster segment's preferences and behaviour. Results will emphasise data quality and algorithm choice. This research will conclude with insights on how online retailers can improve customer satisfaction and business performance through customer segmentation and big data analytics. The study found better segmentation methods that allow online retailers to use big data analytics to segment their customers more precisely. Customer insights are better with behavioral, demographic, and real-time data. This research shows that big data analytics can transform online retail by allowing businesses to switch from static, one-size-fits-all segmentation models to dynamic, data-driven approaches that better meet digital consumers needs.
Keywords:	Customer Segmentation, Big Data Analytics, Online Retail Industry, Machine Learning, Clustering Algorithms
Source:	Journal of Theoretical and Applied Information Technology 30^th November 2024 -- Vol. 102. No. 22-- 2024
Full Text

Title:	MACHINE AND DEEP LEARNING MODELS FOR MULTI-CLASS SENTIMENT CLASSIFICATION
Author:	LAMIAA A. GAAFAR, ATEF Z. GHALWASH, ALIAA A. YOUSSIF, HAITHAM A. GHALWASH
Abstract:	Nowadays, Artificial Intelligence (AI) is renowned for embedding human-like intelligence in computers, enabling them to mimic human behavior. A pivotal domain within AI is recommendation frameworks, which aid users by suggesting various choices, thereby facilitating optimal decision-making in contexts like purchasing items, selecting healthcare services, movies, etc. This paper introduces classification based on sentiment analysis, aimed at extracting opinions from user reviews. The analysis employs eight models: five machine learning models— Extreme Gradient Boosting (XGBoost), Naïve Bayes (NB), Logistic Regression (LR), Support Vector Machine (SVM), and Random Forest (RF); two deep learning models—Long Short-Term Memory (LSTM) and a distilled version of Bidirectional Encoder Representations from Transformers (DistilBERT) transformer model; and a proposed model integrates Convolutional Neural Networks (CNNs) and Feedforward Neural Networks (FFNNs), alongside the Mamdani Fuzzy System. Notably, the LSTM model demonstrates superior performance, especially attributed to its efficacy in processing shorter sentences, typically ranging from 15 to 20 words as in the used data set, thus slightly outperforming the DistilBERT transformer model in this context. A comparative analysis between 3-class (positive, negative, and neutral) and 4-class (strongly positive, positive, negative, and neutral) classifications reveals the LSTM predominance across all models. Notably, the Long Short-Term Memory (LSTM) model excels in the 3-class sentiment classification, achieving an accuracy of 0.99, precision of 0.99, recall of 0.99, and an F1 score of 0.99 after oversampling.
Keywords:	Natural Language Processing, Sentiment, Classification, Deep Learning, Oversampling
Source:	Journal of Theoretical and Applied Information Technology 30^th November 2024 -- Vol. 102. No. 22-- 2024
Full Text

Title:	IMPROVING SECURITY IN INTELLIGENT SYSTEMS: HOW EFFECTIVE ARE MACHINE LEARNING MODELS WITH TF-IDF VECTORIZATION FOR PASSWORD-BASED USER CLASSIFICATION
Author:	BOUMEDYEN SHANNAQ
Abstract:	This research assesses the practicability of machine learning models in classifying consumers according to their passwords with the help of TF-IDF, which depicts exclusive password features. The purpose of the study is to eradicate the weakness of the current EPSB algorithm in its synthesis of electronic personal behavior. Our goal will be to define those models that have strengthened the existing methods of password-based authentication. In the second step, we transformed a data set of anonymized passwords to the transformation where each was converted into statistical feature vectors using TF-IDF and tested six models of machine learning. Specific well-known algorithms used in the course of the study were support vector machines (SVM), random forests, Naïve Bayes, K-nearest neighbor (KNN), logistic regression, and decision trees. This cross-validation made me conclude that Naive Bayes outcompeted all the other models in terms of a greater weighted average precision of 96.38%, which was higher than the other two models: the SVM model equal to 91.64% and the logistic regression model equal to 91.52%. With regards to accuracy, KNN got 79.48%, Decision Tree got 77.55%, while Random Forest recorded the lowest value of the four techniques at 71.26%. These results provide a more profound comprehension for the development of an extended password-based authentication scheme using an advanced machine learning approach.
Keywords:	Password Classification, Machine Learning, TF-IDF Vectorization, Support Vector Machine, Random Forest, Naive Bayes-Nearest Neighbors, Logistic Regression, Decision Tree
Source:	Journal of Theoretical and Applied Information Technology 30^th November 2024 -- Vol. 102. No. 22-- 2024
Full Text

Title:	SIMPLE GENETIC ALGORITHM BASED RANDOM TESTING FOR REDUCING FAULTY TEST CASES
Author:	D. SREE LAKSHMI, P. BHARAT SIVA VARMA, K. ASHOK KUMAR, DR.Y. SUMANTH, JANJHYAM VENKATA NAGA RAMESH, G. N. SOWJANYA
Abstract:	The goal of software testing is to identify software flaws. Software testing is the process of confirming that a program works as intended. Test inputs are generated at random from the software's input space during random testing. to consistently provide test instances that are random and have some similarities. We will provide a method for minimizing the errors based on the best test cases produced by directed random testing in order to get around these problems. Using the random testing model as a basis, we will create an effective random testing test case in the suggested approach. The Simple Genetic Algorithm (SGA) will be used in this study to create the best inputs, minimizing both equivalent and illicit inputs. SGA makes use of the test case coverage metrics to lessen fault proneness. By merging the old input with the present one, our suggested approach will reduce the input space and improve scalability and efficacy in the software testing age.
Keywords:	SGA, Software testing, Coverage, RT
Source:	Journal of Theoretical and Applied Information Technology 30^th November 2024 -- Vol. 102. No. 22-- 2024
Full Text

Title:	DO TECHNOLOGY-RELATED INVESTMENTS HAVE A REMARKABLE IMPACT ON THE SCALABILITY AND PRODUCTIVITY OF THE FINANCIAL INDUSTRY?
Author:	ROCHANIA AYU YUNANDA, TOTO RUSMANTO, NURIL KUSUMAWARDANI SOEPRAPTO PUTRI, MOHAMMAD ALI TAREQ
Abstract:	Digital transformation makes investing in information and communications technology essential for all industries including the financial industry. In the Indonesian banking sector as part of the financial industry, technology-driven banking services were found to increase both banking transactions and customer numbers. This study investigates whether technology-related investments significantly enhance the scalability and productivity of commercial banks in Indonesia. This is quantitative research. The research employs a panel data regression of 80 observed data during the period 2019-2023. This study found that spending amount on technology-related investment has an impact on scalability and productivity. Specifically, technology-related investment increase revenue growth and productivity (sales per employee). The research findings have both academic and practical impacts. The results could encourage the bankers and decision makers to determine appropriate technological investment strategies. Investors can also consider highly digitalized banks as valuable investment. This research extends the current literature on the relationship between digitalization and banking performance, with a focus on commercial banks in Indonesia. This study offers two other proxies to assess banking performance: scalability and productivity, as technological advances aim to enhance company growth and productivity.
Keywords:	Technology, Software, Banking, Productivity, Scalability, Indonesia
Source:	Journal of Theoretical and Applied Information Technology 30^th November 2024 -- Vol. 102. No. 22-- 2024
Full Text

Title:	AN INNOVATIVE MACHINE LEARNING FRAMEWORK FOR PHONOCARDIOGRAPHY (PCG) USING MFCC AND DEEP EXTREME LEARNING MACHINE (DELM)
Author:	ABDULLAH ALTAF, HAIRULNIZAM MAHDIN, AWAIS MAHMOOD, ABDULREHMAN ALTAF
Abstract:	Cardiovascular Diseases (CVDs) are a significant global cause of mortality, necessitating effective diagnostic techniques. Phonocardiography (PCG) is among the fundamental methods used to analyze heart sounds to detect human heart-related abnormalities. However, in an environment where state-of-the-art PCG equipment is not available, a Machine Learning (ML) based solution can serve as a reliable alternative. However, the main challenges faced by ML-based PCG systems, are the unavailability of balanced and unbiased datasets, the vanishing and exploding gradient a well-known Deep Learning (DL) issue, and inappropriate feature extraction techniques, which often compromise the accuracy and reliability of ML-based PCG systems. This study introduces a novel Deep Extreme Learning Machine (DELM) and Mel-Frequency Cepstral Coefficients (MFCC) based PCG framework for CVD diagnosis. The proposed framework uniquely addresses the above mentioned challenges. The proposed model achieves a remarkable training accuracy of 98.46 % and a test accuracy of 86.80 %, using the Heartbeat Sound dataset with five classes and after class aggregation and dataset normalization the proposed model achieved training accuracy 99.52 % and a test accuracy of 92.30 % demonstrating its potential in PCG diagnostics. This framework represents a significant advancement in ML-based PCG systems for automating heart sound analysis and contributing to improved cardiac healthcare, especially in resource-limited settings.
Keywords:	Cardio Vascular Disease (CVDs), MFCC, Machine Learning, Deep Extreme Learning Machine (DELM), Heart Disease
Source:	Journal of Theoretical and Applied Information Technology 30^th November 2024 -- Vol. 102. No. 22-- 2024
Full Text

Title:	PROGRESS TRANSFORMER ON ALZHEIMER’S DISEASE PROBABILITY FINDER FROM MILD COGNITIVE IMPAIRMENT
Author:	SARANYA RATHINAM, KRISHNAN NALLAPERUMAL, KALIDASS SUBRAMANIAM
Abstract:	The prodromal phase of Alzheimer's disease (AD) is called mild cognitive impairment (MCI). Effective treatments depend on identifying MCI patients who have a high chance of transforming AD. This paper proposes a temporal magnetic resource imaging (MRI) slice feature analysis using transformers to predict the chance of AD. The proposed Progress Transformer (ProgTransAD) model finds the relative changes in the MRI slices with the help of encoding the convolutional backbone feature maps and their corresponding cosine similarity. The proposed deep learning approach forecasts whether someone would develop Alzheimer's disease (AD) after receiving a diagnosis of MCI with three years of analysis. The performance of this unique deep learning network which can accurately diagnose AD progression is analyzed using the Alzheimer's Disease Neuroimaging Initiative (ADNI-1) dataset and this ProgTransAD achieves 94% accuracy one year ahead.
Keywords:	Alzheimers disease, Mild Cognitive Impairment, Neuroimaging, Magnetic Resource Imaging, ProgTransAD
Source:	Journal of Theoretical and Applied Information Technology 30^th November 2024 -- Vol. 102. No. 22-- 2024
Full Text

Title:	UNPRECEDENTED HARMONY SEARCH OPTIMIZATION-BASED LEACH ROUTING PROTOCOL (UHSO-LRP) FOR ENHANCING WIRELESS BODY AREA NETWORKING (WBAN) LIFETIME
Author:	S.VEERARATHINAKUMAR , Dr. B.DEVANATHAN
Abstract:	Wireless Body Area Networks (WBANs) play a crucial role in healthcare applications by enabling continuous and non-invasive monitoring of patients' vital signs and health data through wearable devices. These networks transmit real-time data from wearable sensors to a central monitoring system, enabling remote patient monitoring and timely medical intervention. Routing in WBANs is a critical aspect as it involves the selection of optimal paths for data transmission among various wearable devices and the central node. Traditional routing protocols face challenges in WBANs due to the characteristics of the human body, such as dynamic channel conditions, varying distances between devices, and energy constraints of wearable devices. Conventional routing protocols may not efficiently handle these issues, leading to suboptimal performance, increased energy consumption, and limited network lifetime. The proposed work, Unprecedented Harmony Search Optimization-Based LEACH Routing Protocol (UHSO-LRP), aims to address the issues faced in routing within WBANs. It introduces a hybrid approach that combines the Harmony Search Optimization (HSO) algorithm with the Low-Energy Adaptive Clustering Hierarchy (LEACH) routing protocol. The working mechanism of UHSO-LRP involves the utilization of the HSO algorithm to optimize the selection of cluster heads and the routing of data packets. The HSO algorithm introduces self-optimizing capabilities, allowing the network to dynamically adjust its configuration to changing conditions. UHSO-LRP effectively manages energy consumption and prolongs the network lifetime by optimizing the cluster head selection and data routing. Simulations are conducted to evaluate the performance of UHSO-LRP compared to conventional routing protocols. The simulations analyze key performance metrics, such as network lifetime, energy efficiency, packet delivery ratio, and latency. The results demonstrate that UHSO-LRP outperforms traditional routing protocols, showcasing significant network stability and energy utilization improvements.
Keywords:	Wireless Body Area Networks (WBANs), Routing Protocol, Harmony Search Optimization (HSO), Low Energy Adaptive Clustering Hierarchy (LEACH), Wireless Sensor Networks (WSN), Healthcare Monitoring
Source:	Journal of Theoretical and Applied Information Technology 30^th November 2024 -- Vol. 102. No. 22-- 2024
Full Text