Journal of Artificial Intelligence and Metaheuristics
Journal DOI
2833-5597ISSN (Online)
Accurate prediction of electricity consumption is a critical requirement for improving operational efficiency, enhancing grid reliability, and supporting sustainability objectives in urban power distribution systems, particularly in regions experiencing steady population growth and increasing demand pressure. Motivated by the limitations of conventional statistical and physics-inspired forecasting approaches, as well as the strong sensitivity of deep learning architectures to hyperparameter configuration, t his s tudy p roposes a robust data-driven framework that integrates deep learning with advanced metaheuristic optimization for high-precision short-term electricity consumption forecasting. The main contribution of this work lies in the systematic development and evaluation of hybrid metaheuristic–Bidirectional Long Short-Term Memory (BiLSTM) models, in which multiple state-of-the-art optimization algorithms are employed to tune model hyperparameters. Particular emphasis is placed on the integration of the Ninja Optimization Algorithm with BiLSTM (NijOA + BiLSTM), which is designed to effectively navigate complex, high-dimensional hyperparameter search spaces encountered in deep learning–based load forecasting tasks. Baseline experiments demonstrate that BiLSTM outperforms other deep learning models, including Artificial Neural Network (ANN), Recurrent Neural Network (RNN), Long Short-Term Memory (LSTM), and Gated Recurrent Unit (GRU), achieving a baseline Root Mean Squared Error (RMSE) of 0.0964 and a coefficient of determination (R2) of 0.854. These results confirm t he a dvantage o f b idirectional t emporal l earning in capturing the nonlinear and time-dependent characteristics of electricity consumption recorded at high temporal resolution from SCADA systems. Following metaheuristic optimization, the NijOA + BiLSTMmodel delivers a substantial improvement in predictive performance. The optimized configuration reduces RMSE to 0.0038, Mean Squared Error (MSE) to 1.45 × 10−5, and Mean Absolute Error (MAE) to 0.00019, while increasing the correlation strength to r = 0.973 and the explanatory power to R2 = 0.97. Comparative analysis across different optimization strategies further confirms t he s uperiority o f t he NijOA + BiLSTM hybrid model over alternative configurations, including WAO + BiLSTM, BBO + BiLSTM, GA + BiLSTM, SFS + BiLSTM, DE + BiLSTM, and JAYA + BiLSTM. The implications of these findings are significant for real-world urban electricity distribution applications. The proposed framework enables highly accurate and reliable short-term electricity consumption forecasting, making it well suited for deployment within smart grid and distribution management systems. Such predictive capability can support informed operational decision-making, improve demand-side management strategies, reduce uncertainty in short-term planning, and contribute to the long-term sustainability and resilience of urban power distribution networks.
Read MoreDoi: https://doi.org/10.54216/JAIM.110101
Vol. 11 Issue. 1 PP. 01-28, (2026)
The growing heterogeneity of cardiovascular disease presentations poses significant c hallenges for clinical decision support systems, particularly in identifying patient similarities and developing robust predictive models capable of supporting personalized treatment strategies, which motivates the need for advanced data-driven frameworks that can jointly exploit unsupervised learning, deep learning, and intelligent optimization. In this study, we propose a comprehensive hybrid framework that integrates unsupervised patient clustering with deep learning classification, enhanced through Fitness Greylag Goose Optimization (FGGO), where clustering is first employed to uncover latent patient subgroups and inform downstream learning, followed by the use of a Deep Learning Framework Distilled by Gradient Boosting Decision Trees (DeepGBM) as the core predictive model, and finally optimized via FGGO for automated hyperparameter tuning. The primary contribution of this work lies in the design of an FGGO-optimized DeepGBM framework that systematically improves learning stability, feature interaction modeling, and predictive robustness, while also providing a rigorous comparative evaluation against other state-of-the-art metaheuristic optimizers, including Particle Swarm Optimization (PSO), Grey Wolf Optimizer (GWO), Dipper Throated Optimization (DDTO), and Multiverse Optimization (MVO). Experimental results demonstrate that, at the baseline stage without optimization, DeepGBM achieves an accuracy of 0.9032, sensitivity of 0.8824, specificity of 0.9195, and F-score of 0.8889, indicating strong but improvable performance on heart disease patient data. After metaheuristic optimization, the proposed FGGO + DeepGBM model exhibits a substantial performance enhancement, reaching an accuracy of 0.9795, sensitivity of 0.9747, specificity of 0.9831, positive predictive value of 0.9776, negative predictive value of 0.9809, and an F-score of 0.9761, consistently outperforming PSO + DeepGBM, GWO + DeepGBM, DDTO + DeepGBM, and MVO + DeepGBM across all evaluation metrics. These results highlight the robustness and convergence consistency of FGGO-based optimization and confirm i ts e ffectiveness in navigating complex hyperparameter search spaces. The implications of this work extend to clinical practice and intelligent healthcare systems, as the proposed framework offers a reliable and scalable solution for patient stratification and heart disease prediction, supporting more accurate, interpretable, and data-driven clinical decision-making while paving the way for future integration into personalized and precision medicine applications.
Read MoreDoi: https://doi.org/10.54216/JAIM.110102
Vol. 11 Issue. 1 PP. 29-57, (2026)
The growing prevalence and clinical complexity of Obsessive–Compulsive Disorder (OCD) motivate the need for reliable, data-driven decision-support systems capable of improving diagnostic accuracy and robustness beyond traditional assessment methods. In this study, we propose an optimized deep learning framework that integrates a Deep Learning framework distilled by Gradient Boosting Decision Trees (DeepGBM) with a novel metaheuristic optimizer, the Ninja Optimization Algorithm (NiOA), to enhance OCD-related classification using structured demographic and clinical data. The main contribution of this work lies in the design of a unified optimization pipeline in which NiOA is employed for automated hyperparameter tuning of DeepGBM, and in the comprehensive comparison of this approach against baseline deep learning models and alternative metaheuristic optimizers, including Multiverse Optimization (MVO), Bat Algorithm (BA), and Particle Swarm Optimization (PSO). Experimental evaluation demonstrates that, at the baseline stage, DeepGBM outperforms Artificial Neural Networks (ANN), Convolutional Neural Networks (CNN), and Bidirectional Long Short-Term Memory networks (BiLSTM), achieving an accuracy of 0.8970 and an F-score of 0.8935. Following optimization, the proposed NiOA+DeepGBM framework achieves substantial performance gains, reaching an accuracy of 0.9779, sensitivity of 0.9763, specificity of 0.9793, and an F-score of 0.9770, consistently surpassing MVO+DeepGBM, BA+DeepGBM, and PSO+DeepGBM across all evaluation metrics. These results confirm the superior capability of NiOA in navigating complex hyperparameter spaces and enhancing both predictive accuracy and generalization. The implications of this work are significant for intelligent mental health assessment, as the proposed NiOA-optimized DeepGBM model offers a robust, clinically relevant decision-support tool that can assist clinicians in improving diagnostic reliability, reducing uncertainty, and supporting the development of scalable, AI-driven mental healthcare systems.
Read MoreDoi: https://doi.org/10.54216/JAIM.110103
Vol. 11 Issue. 1 PP. 58-86, (2026)
The rapid increase in liver disease prevalence worldwide, particularly in developing regions, necessitates accurate and reliable diagnostic systems capable of supporting early clinical decision-making based on routine laboratory data. Traditional diagnostic approaches and unoptimized machine learning models often struggle to fully capture the complex, nonlinear relationships among biochemical liver indicators, leading to suboptimal predictive reliability. Motivated by these challenges, this study proposes a human-inspired metaheuristic optimization framework that integrates the iHow Optimization Algorithm (iHOW) with the Extreme Gradient Boosting model (XGBoost) to enhance liver disease prediction performance. The main contribution of this work lies in the development of an optimized diagnostic pipeline that systematically tunes XGBoost hyperparameters using iHOW and rigorously benchmarks its effectiveness against established metaheuristic optimizers, including Genetic Algorithm (GA), Particle Swarm Optimizer (PSO), Grey Wolf Optimizer (GWO), and Greylag Goose Optimization (GGO). Experimental evaluation is conducted on a clinically sourced liver disease dataset using multiple diagnostic metrics. In the baseline stage, the unoptimized XGBoost model achieves an accuracy of 0.921875, sensitivity of 0.920245399, specificity of 0.923566879, and F-Score of 0.923076923. After hyperparameter optimization, the proposed iHOW+XGBoost framework demonstrates substantial performance enhancement, attaining an accuracy of 0.983696458, sensitivity of 0.983391608, specificity of 0.984012066, and F-Score of 0.983965015, outperforming GA+XGBoost, PSO+XGBoost, GWO+XGBoost, and GGO+XGBoost across all evaluated metrics. These results confirm the effectiveness of human-inspired optimization in navigating complex hyperparameter search spaces and improving diagnostic robustness. The findings of this study highlight the practical implications of integrating advanced metaheuristic optimization with ensemble learning models, offering a highly accurate, reliable, and scalable decision-support framework that can be leveraged for early liver disease screening and extended to other medical diagnostic and predictive healthcare applications.
Read MoreDoi: https://doi.org/10.54216/JAIM.110104
Vol. 11 Issue. 1 PP. 87-116, (2026)
Gamma–hadron discrimination remains a fundamental challenge in very-high-energy gamma-ray astronomy due to the strong overlap between gamma-ray–initiated and hadron-induced air showers recorded by imaging atmospheric Cherenkov telescopes, particularly at low energies where background contamination is severe. Traditional cut-based and non-optimized machine learning approaches often struggle to fully exploit the nonlinear and correlated nature of Cherenkov image parameters, leading to suboptimal background suppression and reduced telescope sensitivity. To address these limitations, this paper proposes a unified deep learning and metaheuristic optimization framework that combines an enhanced attention-based long short-term memory network (EALSTM) with advanced optimization strategies. In particular, a novel Adaptive Balanced Greylag Goose Optimization algorithm (ABGGO) is employed to jointly perform feature selection and hyperparameter optimization, enabling effectiveexploration–exploitation balancing while preserving physically meaningful feature representations. The proposed ABGGO+EALSTM framework is systematically evaluated against baseline deep learning models, including artificial neural networks (ANN), convolutional neural networks (CNN), and standard long short-term memory networks (LSTM), under identical experimental conditions. Experimental results on a Monte Carlo–generated Cherenkov telescope dataset demonstrate clear and consistent performance gains at every stage of the analysis. In the baseline evaluation stage, EALSTM achieves an accuracy of 0.9294 and an F-score of 0.9266, outperforming ANN, CNN, and LSTM. Following metaheuristic optimization, the proposed ABGGO+EALSTM model attains a peak accuracy of 0.9718, sensitivity of 0.9694, specificity o f 0 .9740, a nd F-score o f 0 .9705, representing absolute improvements exceeding 4% over the baseline EALSTM configuration and outperforming GA+EALSTM, GWO+EALSTM, and PSO+EALSTM variants. These results demonstrate that integrating attention-based deep learning with adaptive metaheuristic optimization significantly enhances gamma–hadron discrimination, leading to improved background suppression and signal retention. The proposed framework offers a scalable and robust solution for current and next-generation Cherenkov observatories, with strong potential for real-time event filtering, multi-telescope analysis, and future deployment on real observational data.
Read MoreDoi: https://doi.org/10.54216/JAIM.110105
Vol. 11 Issue. 1 PP. 117-146, (2026)