Predicting Primary Biliary Cholangitis Stages Using Machine Learning with Automated Hyperparameter Optimization and Recursive Feature Elimination

perRICESTJournal of Information Systems and Telecommunication (JIST) 2322-14372345-27732025-111351165176articlePredicting Primary Biliary Cholangitis Stages Using Machine Learning with Automated Hyperparameter Optimization and Recursive Feature EliminationArman Rezasoltaniarmanrezasoltani@ut.ac.ir1Amir Mohammad Khaniamir.mo.khani@ut.ac.ir2Ali Husseinzadeh Kashana.kashan@modares.ac.ir3Shahram AgahAgah.sh@iums.ac.ir4Fatemeh AgahFatemeh.agah@student.adelaide.edu.au5This research used modern machine learning ways to predict the stages of primary biliary cholangitis using data from the Mayo Clinic trial. The research aims to obtain high prediction accuracy while representing balanced evaluation metrics. Important techniques include automated hyperparameters optimization with Optuna and Recursive Feature Elimination to improve model performance. Pre-processing included handling missing values, encoding of categorical features, and addressing class imbalances using SMOTE. A total of twelve machine learning algorithms are evaluated with ensemble-based models such as CatBoost and Extra Trees producing much better results. Evaluation metrics take into account all model predictions, including accuracy, precision, recall, F1 score, and ROC-AUC for performing balanced and interpretative evaluations of performances critical for imbalanced datasets. This endeavor includes clinical and laboratory information illustrating the prospect of machine learning in advancing therapeutic diagnosis, emphasizing the rigor and robustness in evaluation laid groundwork for future research to encompass even more generalizable and robust diagnostic tools.http://jist.ir/Article/49352Primary Biliary Cholangitis Machine Learning Recursive Feature Elimination Optuna Imbalanced Data.perRICESTJournal of Information Systems and Telecommunication (JIST) 2322-14372345-27732025-111351177188articleResolving Class Imbalance in Medical Classification: Technique Comparison and Performance EvaluationAbdallah Maitiabdallah.maiti@uhp.ac.ma1Mohamed Haninimohamed.hanini@uhp.ac.ma2Abdallah Abardaabdallah.abarda@uhp.ac.ma3The problem of unbalanced data is a common one in medical diagnostics. This problem can reduce the accuracy of classification models and affect the validity of results. The aim of our paper is to compare several techniques for correcting class imbalances in medical datasets and to evaluate the impact of these techniques on machine learning performance. In our paper, we used an imbalanced dataset to train a convolutional neural network (CNN) model. We then tested correction techniques such as sampling and cost-sensitive learning. Finally, we used recall, precision, accuracy and F1 score to evaluate the model's performance. The results show that the use of correction techniques led to a significant improvement in the performance of the classification model. The cost-sensitive learning technique gave the best results, particularly for the detection of minority classes. This method increased the weight of classification errors associated with minority classes, thus improving the detection of critical cases. The results of this study underline the importance of dealing with imbalances in the data to improve the performance of classification models in the medical field. The use of methods such as cost-sensitive learning not only improves model performance, but also enables more reliable decisions to be made, which is essential for ensuring more accurate diagnoses and better quality of care.http://jist.ir/Article/49725Data Imbalance Techniques for Resolving Data Class Imbalance Oversampling Cost-Sensitive learning Convolutional Neural Networks Classification Model Performance Medical Diagnostics.perRICESTJournal of Information Systems and Telecommunication (JIST) 2322-14372345-27732025-111351189209articleEnhancing IoT Security: A Hybrid Deep Learning-Based Intrusion Detection System Utilizing LSTM, GRU, and Attention Mechanisms with Optimized Hyperparameter TuningHeshamt Asadiheshmat.asadi94@yahoo.com1Mahmood AlborziMahmood_alborzi@yahoo.com2Hessam ZandhessamiZandhessami@srbiau.ac.ir3Increasing complexity and volume of threats being created and targeted at cybersecurity for the IoTs necessitate the deployment of powerful IDSs. This paper offers an innovative intrusion detection system for IoTs networks based on deep learning. The new IDS employs the Long Short-Term Memory and Gated Recurrent Unit models’ strengths and an Attention Mechanism. First, the new IDS seeks to enhance the model’s ability to determine critical features in a vast amount of data streams and hence improve the ability to find potential cyber threats with high accuracy. The methodological framework used in a simulation and practical experiment setting was intended to recognize the unique nature of IoTs situations. therefore, used a hybrid algorithm optimization strategy, namely Differential Evolution and Harmony Search, to optimize the model due to the extensive hyperparameter space to get the best performance results. The results obtained superior accuracy, precision, recall, and F1 measures reaching 99.87 percent, 99.84 percent, 99.85 percent, and 99.85 percent is better than the performance measures achieved by existing models. Therefore, a deep learning-based hybrid IDS confirmed the research hypothesis that this could provide the necessary and effective cybersecurity for the IoTs. It is vital to note that this paper has contributed to the research topic by showing the potential of advanced neural architectures and strategic optimization tools to address the massive and sophisticated IoTs cybersecurity issues. Future research will be addressing whether these models can be applied in more IoTs settings and whether their real-time efficiency can be improved.  http://jist.ir/Article/49180Intrusion Detection System in Internet of Things Attention Mechanism in Deep Learning algorithm Differential Evolution Harmony SearchperRICESTJournal of Information Systems and Telecommunication (JIST) 2322-14372345-27732025-111351210231articleTowards Energy-efficient Cloud Computing: A Review of Network-Aware VM Placement ApproachesAli Baydounal.baydoun@bau.edu.lb1Ahmed S ZekriAhmed.zekri@alexu.edu.eg2Cloud data centers (CDCs) have witnessed significant growth to meet the increasing demands of modern applications. However, this expansion has raised concerns regarding the environmental impact, energy requirements, and electricity costs associated with data centers. The network infrastructure, serving as the communication backbone of these data centers, plays a crucial role in their scalability, performance, cost, and, most importantly, energy consumption. This review provides meaningful perspectives and valuable insights into the state-of-the-art research regarding the problem of virtual machine placement (VMP), focusing on the network-aware energy efficiency aspects of data centers. It provides an overview of VM placement and presents a comprehensive survey of prominent VM placement algorithms from the existing literature. Additionally, a thematic taxonomy of network-aware algorithms is introduced, highlighting the key energy consumption metrics and presenting a new classification of VMP algorithms that considers datacenter network (DCN) topology, traffic patterns, communication patterns, and energy reduction strategies. Besides addressing pertinent research questions in this domain, this review summarizes the findings and suggests potential avenues for future research, guiding researchers in designing and implementing more effective and efficient network-aware VM placement algorithms that optimize energy consumption, improve network performance, and minimize migration costs.http://jist.ir/Article/49070Cloud computing VM placement network-aware Energy-efficient Network architectureperRICESTJournal of Information Systems and Telecommunication (JIST) 2322-14372345-27732025-111351232242articleSimulation Based Economical Approach for Detecting Heart Disease Earlier from ECG DataMd. Obaidur Rahamanobaidur@eub.edu.bd1Mohammod Abul Kashemdrkashemll@duet.ac.bd2Sovon Chakrabortysovon.chakraborty@ulab.edu.bd3Shakib Mahmud Diptodiptomahmud2@gmail.com4Cardiovascular diseases present significant challenges to public health in developing countries. The high costs of traditional treatments and the limited availability of specialized medical equipment contribute to these challenges. Current diagnostic methods often rely on specific electrocardiogram (ECG) parameters, which may not capture the nuanced complexities necessary for accurate diagnosis. To address these issues, our study proposes an innovative solution: an accessible and cost-effective ECG monitoring system. This system not only captures electrical signals from the heart but also translates them into numerical values using advanced modulation techniques. A trained deep learning model then analyzes this data to accurately identify any potential complications or confirm a healthy cardiac state. Our approach also allows for remote diagnosis and treatment. By utilizing an MQTT server, ECG data can be efficiently transmitted to experts for evaluation and intervention when necessary. Our meticulously fine-tuned Artificial Neural Network (ANN) architecture has achieved an impressive accuracy of 95.64%, surpassing existing methodologies in this field. Designed with resource-strapped regions in mind, our system offers a lifeline to rural areas lacking access to medical professionals and advanced equipment. Its affordability ensures that even individuals with limited financial means can benefit from timely and accurate cardiac monitoring, potentially saving lives and reducing the burden of cardiovascular diseases in underprivileged communities.http://jist.ir/Article/47968Artificial Neural Network (ANN) Cardiovascular D isease Electrocardiogram Heart Disease Modulation Techniques MQTT ServerperRICESTJournal of Information Systems and Telecommunication (JIST) 2322-14372345-27732025-111351243255articleEnhancing Computational Offloading for Sustainable Smart Cities: A Deep Belief Network ApproachKaebeh Yaeghoobiyaeghoobi@kntu.ac.ir1Mahsa Bakhshandeh Nmahsabakhshandeh1995@gmail.com2The use of mobile devices with limited processing power has surged in recent years, alongside the expansion of cloud and fog computing across various sectors. These devices can handle small to medium computing tasks, but they fall short when it comes to large-scale processes, making computational offloading a crucial solution. Cloud computing and fog computing provide an effective platform for offloading tasks from mobile devices. However, critical real-time applications necessitate a near-edge approach to managing the computational load. Significant challenges exist in optimizing response times for effective offloading in cloud computing. This research introduces a framework for predicting response times using Deep Belief Network (DBN) learning to enhance offloading performance. Implementing a DBN aims to minimize response times and resource consumption, thereby improving the overall efficiency of offloading processes. The framework is designed to predict response times accurately, ensuring timely completion of tasks and efficient use of resources. Simulation results using multiple models show that the use of DBN significantly reduces processing, response, and offloading times compared to other algorithms. Consequently, the DBN algorithm proves to be more efficient in predicting response times and enhancing offloading performance. By leveraging the capabilities of DBN, this framework provides a promising solution for optimizing computational offloading in cloud computing environments. This enhances the performance of mobile devices and ensures the reliability and efficiency of real-time applications, direct the way for more advanced and responsive computing technologies.http://jist.ir/Article/48580Computational Offloading Cloud Computing Deep Belief Network Response Time Resource Management Sustainable Smart Cities Real-time ManagementperRICESTJournal of Information Systems and Telecommunication (JIST) 2322-14372345-27732025-111351256265articlePSO-Optimized Power Allocation in NOMA-QAM for Beyond 5G: A CFD and MFD AnalysisJaspreet Kaurjaspreetsweetangel@gmail.com1This paper proposes a power allocation method based on particle swarm optimization (PSO) to enhance spectrum sensing performance in downlink Non Orthogonal Multiple Access (NOMA) systems employing high-order Quadrature Amplitude modulation (QAM) modulation for beyond 5G networks. By intelligently adjusting user power levels, the proposed approach significantly improves detection reliability while maintaining stringent false alarm constraints, even under challenging low-SNR conditions. The goal is to enhance spectrum sensing performance by maximizing the probability of detection (Pd) while maintaining a constrained probability of false alarm (Pf). Cyclostationary Feature Detection (CFD) and Matched Filter Detection (MFD) techniques are applied to evaluate detection performance under varying Signal to noise ratio (SNR) conditions. Simulation results demonstrate that the optimized framework not only strengthens detection performance particularly for high order QAM but also enhances overall system responsiveness.  Also CFD surpasses MFD in higher SNR scenarios due to its ability to exploit cyclic features of modulated signals, which are preserved even in moderately noisy environments. The integration of PSO further enhances system performance, offering a practical and scalable solution for next-generation Internet of Things (IoT)-enabled spectrum sharing environments.http://jist.ir/Article/48051Non Orthogonal Multiple Access (NOMA) Matched Filter Detection (MFD) CFD PSO Cognitive Radio Networks (CRN) Next Generation Networks (NGN)