Journal of Computing Theories and Applications

Big Data-Driven Health Risk Stratification: A Health Index-Based Approach Using Feature Importance and PySpark

2025-02-13T11:11:18+00:00

Health risk stratification is crucial for preventive healthcare, yet existing models often rely on binary classification generalized disease prediction, neglecting personalized health indicators and graded risk levels. Many studies apply feature selection techniques like Relief and Univariate Selection without quantifying the weighted impact of features. To address these gaps, this study introduces a Big Data-driven Health Index (HI) framework using PySpark for scalable health risk stratification. The HI is computed as a weighted sum of health-related features using SHAP Analysis, XGBoost, Random Forest, and Correlation Analysis. PySpark enables efficient processing of large-scale health data, and individuals are classified into Low and High Risk. Optimal classification thresholds are determined using the Youden Index from the ROC curve to balance sensitivity and specificity. Personalized health recommendations are generated based on risk categories to guide preventive interventions. Performance evaluation reveals that Correlation Analysis achieves 100% precision and 98.90% recall, outperforming other methods. SHAP prioritizes recall but has low precision, while XGBoost and Random Forest improve precision but struggle with recall. By leveraging Big Data techniques with PySpark, this study enhances computational efficiency, scalability, and classification accuracy, addressing prior research limitations and providing a robust data-driven approach to personalized health monitoring.

Aspect-Based Sentiment Analysis on E-commerce Reviews using BiGRU and Bi-Directional Attention Flow

2025-02-12T02:07:48+00:00

Aspect-based sentiment Analysis (ABSA) is vital in capturing customer opinions on specific e-commerce products and service attributes. This study proposes a hybrid deep learning model integrating Bi-Directional Gated Recurrent Units (BiGRU) and Bi-Directional Attention Flow (BiDAF) to perform aspect-level sentiment classification. BiGRU captures sequential dependencies, while BiDAF enhances attention by focusing on sentiment-relevant segments. The model is trained on an Amazon review dataset with preprocessing steps, including emoji handling, slang normalization, and lemmatization. It achieves a peak training accuracy of 99.78% at epoch 138 with early stopping. The model delivers a strong performance on the Amazon test set across four key aspects: price, quality, service, and delivery, with F1 scores ranging from 0.90 to 0.92. The model was also evaluated on the SemEval 2014 ABSA dataset to assess generalizability. Results on the restaurant domain achieved an F1-score of 88.78% and 83.66% on the laptop domain, outperforming several state-of-the-art baselines. These findings confirm the effectiveness of the BiGRU-BiDAF architecture in modeling aspect-specific sentiment across diverse domains.

A Personalized Context-Aware Places of Interest Recommender System

2025-02-12T02:11:30+00:00

This research presents a personalized, context-aware recommender system to suggest Places of Interest (POIs) using a hybrid approach combining Bayesian inference and collaborative filtering. The system explicitly addresses the cold-start problem that new users face and improves recommendation accuracy by considering contextual variables such as user mood, budget, companion, and location. The system collects real-time contextual inputs for new users with no historical data and applies Bayesian inference to generate relevant POI suggestions. As users begin to interact and provide ratings, the system progressively shifts to a collaborative filtering mechanism, leveraging cosine similarity to identify similar users within comparable contexts. The recommender system focuses on three categories of POIs: restaurants, hotels, and landmarks. These locations are retrieved through the Google Maps API, and only mapped locations are considered. The system was implemented on Android devices and evaluated through a user study involving 25 participants from diverse backgrounds, including software developers, IT students, and general users. Evaluation metrics such as normalized Discounted Cumulative Gain (nDCG) and classification accuracy were used to assess recommendation quality. Results demonstrate that the system performs better than traditional methods, with nDCG improvements reaching up to 83 percent. Users reported high satisfaction regarding the recommendations' accuracy, ease of use, and contextual relevance. While the system offers significant improvements, it also has certain limitations. Its dependency on Google Maps data may restrict its scope, and using only four contextual factors limits the system’s adaptability to more complex user preferences. Future enhancements could include additional dynamic contexts such as weather, POI popularity, and time-related trends, as well as integrating more advanced models to increase personalization and flexibility in real-world applications.

An Intelligent Route Planning Approach Using Modified Particle Swarm Optimization for Robot Assisted Minimally Invasive Surgery

2025-03-06T08:10:38+00:00

Minimally invasive surgery offers several advantages, including reduced blood loss, smaller incisions, less pain, and a lower risk of complications than open surgery. This approach enhances patient comfort and supports faster recovery. When guided by optimal path planning, surgical robots can accurately navigate the body to remove malignant tumors with high precision. This study proposes a Modified Particle Swarm Optimization (MPSO) algorithm to determine the optimal path for robotic-assisted minimally invasive surgery targeting brain tumors. The algorithm improves upon standard PSO by modifying the velocity update equation and incorporating an adaptive inertia weight, enhancing convergence speed, global search ability, and solution accuracy. Experimental results show that the proposed MPSO achieves a maximum fitness value of 19.10 in a sparse obstacle environment, outperforming standard PSO and IPSO in quality and in the required number of iterations. The approach effectively balances path efficiency and obstacle avoidance, making it well-suited for complex surgical scenarios. In conclusion, the MPSO-based method provides a reliable and precise solution for robotic surgical navigation, improving outcomes and safety in minimally invasive procedures.

A Comparative Analysis of an Enhanced Hybrid Model for Predicting Dollar Against Naira Exchange Rate Using Deep Learning and Statistical Methods

2025-03-21T06:37:50+00:00

In today’s global economy, accurately predicting foreign exchange rates or estimating their trends correctly is crucial for informed investment decisions. Despite the success of standalone models like ARIMA and deep learning models like LSTM, challenges persist in capturing both linear and nonlinear dynamics in highly volatile exchange rate environments. Motivated by the limitations of these individual models and the need for more robust forecasting tools, this study proposes a hybrid ARIMA-LSTM model that integrates ARIMA’s strength in modeling linear trends with LSTM’s capability to capture nonlinear dependencies, using historical USD/NGN exchange rate data from the Central Bank of Nigeria (CBN) spanning 2001 to 2024. The research hypothesis posits that the hybrid ARIMA-LSTM model will significantly outperform standalone models in forecasting accuracy. By comparing these models against state-of-the-art approaches, the study highlights the advantages of hybridizing statistical and deep learning methods. The findings demonstrate that the hybrid model achieved the lowest Root Mean Squared Error (RMSE) of 2.216 and the highest R² of 0.998, indicating superior forecasting performance. This study fills a critical research gap by demonstrating the effectiveness of hybrid deep learning in financial time series forecasting, providing valuable insights for investors, policymakers, and financial analysts. Future research will extend this work by incorporating the latest dataset and evaluating model robustness during the recent surge in the Naira/Dollar exchange rate from 2023 to 2024.

Adaptive Cyber Defense using Advanced Deep Reinforcement Learning Algorithms: A Real-Time Comparative Analysis

2025-04-06T11:48:08+00:00

Cybersecurity is continuously challenged by increasingly sophisticated and dynamic cyber-attacks, necessitating advanced adaptive defense mechanisms. Deep Reinforcement Learning (DRL) has emerged as a promising approach, offering significant advantages over traditional intrusion detection methods through real-time adaptability and self-learning capabilities. This paper presents an advanced adaptive cybersecurity framework utilizing five prominent DRL algorithms: Deep Q-Network (DQN), Proximal Policy Optimization (PPO), Twin Delayed DDPG (TD3), Soft Actor-Critic (SAC), and Asynchronous Advantage Actor-Critic (A3C). The effectiveness of these algorithms is evaluated within complex, realistic simulation environments using live-streaming data, emphasizing key metrics such as accuracy (AUC-ROC), response latency, and network throughput. Experimental results demonstrate that the SAC algorithm consistently achieves superior detection accuracy (95% AUC-ROC) and minimal disruption to network performance compared to other approaches. Additionally, A3C provides the fastest response times suitable for real-time defense scenarios. This comprehensive comparative analysis addresses critical research gaps by integrating both traditional and novel DRL techniques and validates their potential to substantially improve cybersecurity defense strategies in realistic operational settings.

Integrating Hybrid Statistical and Unsupervised LSTM-Guided Feature Extraction for Breast Cancer Detection

2025-05-03T07:55:59+00:00

Breast cancer is the most prevalent cancer among women worldwide, requiring early and accurate diagnosis to reduce mortality. This study proposes a hybrid classification pipeline that integrates Hybrid Statistical Feature Selection (HSFS) with unsupervised LSTM-guided feature extraction for breast cancer detection using the Wisconsin Diagnostic Breast Cancer (WDBC) dataset. Initially, 20 features were selected using HSFS based on Mutual Information, Chi-square, and Pearson Correlation. To address class imbalance, the training set was balanced using the Synthetic Minority Over-sampling Technique (SMOTE). Subsequently, an LSTM encoder extracted non-linear latent features from the selected features. A fusion strategy was applied by concatenating the statistical and latent features, followed by re-selection of the top 30 features. The final classification was performed using a Support Vector Machine (SVM) with RBF kernel and evaluated using 5-fold cross-validation and a held-out test set. Experimental results showed that the proposed method achieved an average training accuracy of 98.13%, F1-score of 98.13%, and AUC-ROC of 99.55%. On the held-out test set, the model reached an accuracy of 99.30%, precision of 100%, and F1-score of 99.05%, with an AUC-ROC of 0.9973. The proposed pipeline demonstrates improved generalization and interpretability compared to existing methods such as LightGBM-PSO, DHH-GRU, and ensemble deep networks. These results highlight the effectiveness of combining statistical selection and LSTM-based latent feature encoding in a balanced classification framework.

An Empirical Analysis of Injection Attack Vectors and Mitigation Strategies in Redis NoSQL Database

2025-04-24T22:47:17+00:00

The contemporary landscape of data management, marked by an unprecedented scale and velocity of data, has spurred the widespread adoption of NoSQL databases, prioritizing scalability and performance over traditional relational constraints. While offering significant flexibility, this paradigm shift introduces complex cybersecurity challenges, notably query injection vulnerabilities, which are consistently ranked among the top web application security risks. Redis, a leading in-memory key-value store powering critical infrastructure globally, presents a unique security profile due to its architectural design and features like Lua scripting. Despite its prevalence, a comprehensive academic evaluation of Redis injection attack vectors remains understudied. This study addresses this gap by systematically evaluating command and Lua script injection vulnerabilities in Redis version 7.4.1 across controlled configurations: default, password-protected, and ACL-secured environments. We quantify vulnerability risk and empirically validate mitigation strategies by employing a Dockerized testing framework, Python-driven exploit simulations, and CVSS v3.1 scoring. Our findings reveal critical weaknesses in default and permissively configured environments and demonstrate that restrictive Access Control Lists (ACLs), adhering to the principle of least privilege, provide complete mitigation against the specific injection vectors evaluated in our controlled experimental setup. We propose a Redis-specific threat taxonomy and provide empirically validated recommendations for securing Redis deployments, emphasizing layered security controls and proper ACL implementation. This research contributes the first systematic evaluation of modern Redis injection vulnerabilities and highlights the critical importance of security-conscious configurations to protect vital data infrastructure.

A Multilevel Digital Image Thresholding Technique Based on an Enhanced Firefly Algorithm with Neighborhood Attraction

2025-04-26T15:38:38+00:00

Digital image segmentation is essential in image processing, influencing the accuracy of higher-level tasks. Thresholding is widely used, yet identifying optimal threshold values remains challenging. The Firefly Algorithm with Neighbourhood Attraction (FaNA), a metaheuristic approach, is efficient for color image thresholding but underperforms on grayscale images due to suboptimal thresholds. To overcome this, an enhanced version (eFaNA) was developed by integrating a chaotic tent map for population initialization and a Lévy flight-based random walk for improved exploration. eFaNA was compared with FaNA, fuzzy firefly algorithm (FFA), and the standard Firefly Algorithm (FA) in multilevel thresholding of grayscale images. Results demonstrate that eFaNA achieves superior segmentation quality with minimal detail loss, outperforming the others. The average PSNR obtained by eFaNA, FFA, FaNA, and FA was 25.5320 dB, 25.4075 dB, 24.1522 dB, and 24.4506 dB, respectively; average SSIM was 0.8641, 0.8604, 0.8432, and 0.6703; and execution time was 50.5322, 38.7726, 38.7528, and 107.6340 seconds, respectively. This reflects a PSNR improvement of 5.71% over FaNA, 0.49% over FFA, and 4.42% over FA, and an SSIM gain of 2.48% over FaNA, 0.43% over FFA, and 28.92% over FA. While eFaNA lags behind FFA and FaNA in execution time by ~11.8 seconds, it significantly outperforms FA. The performance gain is attributed to the chaotic tent map’s diverse initialization and the Lévy flight’s enhanced search capability. These improvements enable eFaNA to deliver consistently better threshold values and segmentation results. However, its relatively higher computational cost may limit applicability in real-time image processing.