Chronic kidney disease (CKD) is a progressive condition requiring early detection for optimal patient outcomes. This study developed an interpretable machine learning framework using XGBoost with SHapley Additive exPlanations (SHAP) and Local Interpretable Model-agnostic Explanations (LIME) for transparent CKD prediction. We evaluated the approach on two datasets: UAE Tawam Hospital data (n = 491) and UCI CKD data (n = 400). XGBoost with SMOTE optimization achieved 88.4% accuracy (AUC = 0.904) on the hospital dataset and 94.6% accuracy (AUC = 0.948) on the UCI dataset after Rigorous overfitting prevention through conservative hyperparameter ranges and performance monitoring ensured clinical credibility. SHAP analysis identified clinically relevant predictors: eGFRBaseline, HbA1c, and CholesterolBaseline for the hospital cohort, and specific gravity, hemoglobin, and serum creatinine for the UCI cohort. LIME provided complementary patient-level explanations that validated global SHAP patterns. The convergence between global and local interpretability methods confirms model reliability across diverse clinical contexts. This framework addresses the transparency barrier to machine learning adoption in healthcare while maintaining clinically realistic performance levels. The approach provides a foundation for integrating interpretable artificial intelligence into CKD screening and management workflows.

Abstract Early identification of chronic kidney disease (CKD) can prevent progression to renal failure, yet effective screening outside specialist settings remains limited. We present an Internet-of-Things (IoT) platform coupled with a leakage-free machine-learning (ML) pipeline for periodic remote CKD risk scoring. Using a retrospective dataset of electronic health records from 491 adults (21 variables) as a Proof-of-Concept (PoC) to simulate IoT-transmitted patient data, we compare seven classifiers combined with supervised feature selection (FS: Chi-square, ANOVA, Mutual Information) and dimensionality reduction (DR: PCA, UMAP) after imputation, scaling, and class-imbalance handling via model weighting. Performance is estimated with $$10\times $$ repeated, stratified 10-fold cross-validation and a nested threshold selection tailored to screening. To prioritize case detection, we evaluate models at a Screening Operating Point that maximizes recall while maintaining high specificity ( $$\approx 80\%$$ ), accepting the trade-off of modest precision. Under this clinically aligned policy, Logistic Regression (LR) provides the best sensitivity–specificity balance across FS/DR settings, with recalls in the 0.773-−0.843 range and specificities 0.791-−0.803 (e.g., ANOVA FS: recall $$0.843\pm 0.147$$ , specificity $$0.796\pm 0.079$$ , ROC–AUC $$0.880\pm 0.050$$ ; PCA: recall $$0.827\pm 0.137$$ , specificity $$0.803\pm 0.074$$ , ROC–AUC $$0.890\pm 0.045$$ ). At this operating point, precision remains modest (PPV $$\approx 0.33$$ –0.35), reflecting a substantial number of false positives, whereas negative predictive value is consistently high (typically $$\approx 0.96$$ –0.97). For comparison, at a standard F1-optimal operating point, gradient-boosting ensembles increase precision at the cost of recall (e.g., CatBoost + Chi-square: recall $$0.573\pm 0.202$$ ), underscoring the importance of aligning the operating threshold with clinical priorities. Overall, methodological alignment outweighs model complexity for screening utility. Coupled with our scalable IoT architecture, a simple, highly discriminative LR + FS/PCA pipeline—operated under a specificity constraint—offers a low-cost, deployable solution for proactive CKD management.

Chronic Kidney Disease (CKD) represents a significant global health challenge, particularly relevant to Morocco where renal diseases impose a heavy burden on the healthcare system. This study presents an interpretable machine learning framework for CKD prediction using XGBoost models enhanced with SHapley Additive exPlanations (SHAP) and Local Interpretable Model-agnostic Explanations (LIME). Evaluated on a dataset from Tawam Hospital in Abu Dhabi comprising 491 patients with comprehensive demographic, clinical, and biochemical data, the XGBoost model demonstrated exceptional performance. Key predictive biomarkers identified include eGFRBaseline, CholesterolBaseline, and AgeBaseline. These findings not only validate the biological plausibility of the model but also demonstrate its ability to provide accurate, interpretable predictions that can significantly impact healthcare in Morocco. The integration of SHAP and LIME offers clinicians transparent insights into model decisions, enhancing trust and facilitating clinical adoption. This work highlights the transformative potential of interpretable machine learning in nephrology, paving the way for improved patient outcomes through enhanced early detection, risk stratification, and personalized treatment strategies in the Moroccan context.

Financial time-series labeling remains fundamentally limited by three critical deficiencies: temporal rigidity (fixed horizons regardless of market conditions), scale blindness (single-resolution analysis), and correlation-causation conflation. These limitations cause systematic failure during regime shifts. We introduce Adaptive Event-Driven Labeling (AEDL), integrating three core innovations: (1) multi-scale temporal analysis capturing hierarchical market patterns across five time resolutions, (2) causal inference using Granger causality and transfer entropy to filter spurious correlations, and (3) model-agnostic meta-learning (MAML) for adaptive parameter optimization. The framework outputs calibrated probability distributions enabling uncertainty-aware trading strategies. Evaluation on 16 assets spanning 25 years (2000–2025) with rigorous out-of-sample validation demonstrates substantial improvements: AEDL achieves average Sharpe ratio of 0.48 (across all models and assets) while baseline methods average near-zero or negative (Fixed Horizon: −0.29, Triple Barrier: −0.03, Trend Scanning: 0.00). Systematic ablation experiments on a 12-asset subset reveal that selective innovation deployment outperforms both minimal baselines and maximal integration: removing causal inference improves performance to 0.65 Sharpe while maintaining full asset coverage (12/12), whereas adding attention mechanisms reduces applicability to 2/12 assets due to compound filtering effects. These findings demonstrate that judicious component selection outperforms kitchen-sink approaches, with peak individual asset performance exceeding 3.0 Sharpe. Wilcoxon tests confirm statistically significant improvements over Fixed Horizon baseline (p = 0.0024).

The paper presents an analysis of using AIRS and ESN for the prediction of future movements of the EUR/USD exchange rate. Both models, AIRS and ESN, are developed and optimized on historical price data and a number of technical indicators. In order to improve further the prediction of the upward/downturn movement, a stacking ensemble was used to combine the two models using a logistic regression meta-model. On accuracy, AIRS model test set yielded an average of 65.3

This study investigates the enhancement of noise detection in high-frequency ETH/USD trading data by integrating Kalman filter with various machine learning models, optimized using Optuna. The research compares a standalone Kalman filter approach to hybrid models that incorporate machine learning, including Artificial Neural Networks (ANN), Light Gradient Boosting Machines (LightGBM), Random Forests (RF), Convolutional Neural Networks (CNN), Recurrent Neural Networks (RNN), Long Short-Term Memory Networks (LSTM), Bidirectional LSTM (BiLSTM), and Isolation Forests (IF). The models are evaluated using noise reduction metrics such as Mean Squared Error (MSE), Signal-to-Noise Ratio (SNR), and Noise Reduction Ratio (NRR), alongside statistical tests including paired t-tests, Wilcoxon Signed-Rank tests, and ANOVA. Key findings demonstrate that the machine learning-augmented Kalman filters, particularly those using ANN and LightGBM, significantly outperform the traditional Kalman filter in terms of noise suppression. These models achieved lower MSE and higher SNR, indicating better preservation of actionable market signals while reducing irrelevant noise. The statistical tests validate the reliability and effectiveness of these improvements. Although deep learning models like BiLSTM show promise, further optimization is required to match the effectiveness of the leading models. This study contributes to the field of financial time series analysis by demonstrating the potential of combining classical filtering methods with modern machine learning techniques. The results suggest that such integrations can lead to more adaptive and effective noise detection and detection and reduction strategies, potentially improving trading systems in volatile market environments.

Accurately forecasting currency exchange rates is a persistent and significant challenge in computational finance. This study addresses the challenge by introducing an advanced model based on the Artificial Immune Recognition System (AIRS), an algorithm inspired by the adaptive learning of biological immune systems, to predict the directional movement of the EUR/USD pair. While conventional machine learning models are widely used, immune-inspired approaches have been largely unexplored in this domain. Using historical data from May 2002 to July 2024, the proposed model was rigorously optimized through time-series cross-validation and an Evolutionary Algorithm search. On the out-of-sample test set, the optimized model demonstrates strong predictive power, achieving an F1-Score of 0.66 and an ROC AUC of 0.74, results that are competitive with standard machine learning benchmarks. These findings validate AIRS as a robust and scientifically defensible tool for financial forecasting, offering a viable alternative to conventional methods in a highly volatile market.

Precious metals market, such as gold (XAU/USD), exhibit high volatility and significant microstructure noise in their financial time series, which degrade the reliability of algorithmic trading models. While deep reinforcement learning (DRL) has shown strong results in equities and cryptocurrencies, its application to precious metals remains limited by unstable signals and rapid market fluctuations. This study proposes a Kalman-enhanced DRL framework that integrates classical noise filtering with modern neural architectures to improve signal quality and trading performance in highly volatile environments. The methodology applies Kalman filtering to recursively denoise OHLCV price data, which then serves as an input alongside 22 technical indicators to train three state-of-the-art DRL agents: Deep Q-Network (DQN), Proximal Policy Optimization (PPO), and Recurrent PPO (RPPO). Eight years of hourly XAU/USD data (January 2017 to January 2025, N = 47,304) were used for training and evaluation. Models were evaluated on cumulative return, CAGR, Sharpe ratio, maximum drawdown, and volatility. Results demonstrate substantial gains from noise attenuation: PPO with Kalman filtering achieved 80.21% cumulative return (27.1% CAGR, Sharpe 12.10, drawdown-0.48%) compared with raw PPO's 8.70% (3.46% CAGR, Sharpe 0.45, drawdown-12.52%). DQN and RPPO achieved comparable improvements, with 244 to 822% return increases, 88 to 96% drawdown reduction, and up to 29x Sharpe ratio enhancement. Statistical significance was confirmed (p < 0.001 for PPO/RPPO; p < 0.05 for DQN). These findings highlight Kalman-enhanced reinforcement learning as a scalable and robust framework for institutional algorithmic trading, bridging signal processing and artificial intelligence for next-generation adaptive trading systems.

fast technological evolution seen in recent years enhanced the performance and scalability of cloud computing infrastructure and Software-Defined Networking architectures. SDN provides programmability, centralized orchestration, and dynamic resource provisioning, while separating the control and data planes to offer promising architectural paradigm for cloud computing environments. Openness and flexibility expose SDN-based networks to other security concerns, such as large-scale Distributed Denial of Service (DDoS) attacks. This paper introduces a hybrid artificial intelligence (AI) framework for detecting and mitigating DDoS attacks in SDN environments. The framework leverages three complementary approaches: Convolutional Neural Networks (CNN) to capture temporal traffic patterns, Generative Adversarial Networks (GAN) to generate synthetic traffic for dataset augmentation and to enhance anomaly detection, and semi-supervised learning techniques to exploit large amounts of unlabeled traffic data. The proposed system is deployed on a testbed combining OpenDaylight as the SDN controller and Mininet for network emulation, while the AI models are trained and run in Anaconda environment. The network traffic flows are collected, processed into statistical features (i.e., packet rates, entropy values, protocol distribution ratios), and analyzed through the hybrid AI pipeline. Mitigation actions are configured through ODL RESTCONF interface, converting the detection into OpenFlow rules to drop or rate-limit the malicious packets. Experimental evaluation demonstrates that the proposed approach achieves high accuracy detection and robustness to unseen attacks patterns demonstrating the value of applying a hybrid CNN, GAN, Semi-supervised learning approach.

The resurgence of infectious diseases demands innovative forecasting and management strategies. Integrating artificial intelligence (AI) with the internet of things (IoT) offers a groundbreaking approach to epidemic prediction. This study highlights the combined power of AI algorithms and IoT technology in creating sophisticated epidemic models. The authors explore cutting-edge AI methods, like machine learning and deep learning, to analyze vast epidemiological data from IoT devices, including wearables and mobile phones. This integration facilitates early outbreak detection, precise risk evaluation, and prompt interventions. Additionally, they address ethical and privacy issues related to health data, promoting careful information handling. This analysis shows that AI and IoT synergy not only sharpens epidemic forecasting but also boosts public health response efficiency, aiding in epidemic control and enhancing global health security.

Due to the coronavirus crisis, a lot of companies all over the world started a fast digitalization of their business and became more comfortable with the digital world. In this way, a lot of people in the digital domain, including net developers and system engineers, are exposed to different types of risks that they may not completely know or understand. SQL injection is one such vulnerability that can be exploited by attackers to inject malicious data into a database via an application programming interface (API) call request that uses SQL syntax. Cloud computing (IaaS) and network as a service (Naas) provide companies with a wide range of advantages. However, the same advantages can also pose certain vulnerabilities to the organization. In this paper we will focus on SQL injection attacks and make a comparison between different Machine learning algorithms that can be used to detect SQL injection attacks, such as "Naive Bayes" and "K-NN", SVN, and decision tree, we will also use a deep learning algorithm CNN, this comparison will show the potential that machine learning algorithms can offer in terms of cloud security.

This study assesses the effectiveness of machine learning (ML) models in predicting fluctuations in the direction of Bitcoin prices using conventional technical indicators. Given the volatile nature of Bitcoin markets, achieving precise predictions is of utmost importance. The study improves machine learning models including logistic regression (LR), decision trees (DT), random forests (RF), support vector machines (SVM), Xgboost, and artificial neural networks (ANN) by integrating established technical indicators. By employing Optuna for hyperparameter optimization and the 'Time Series Split' technique for cross-validation, the study optimizes the models for time series data. The results demonstrate that ML models, when integrated with technical indicators, significantly outperform predictions based solely on those indicators. Evaluation metrics include accuracy, recall, and the F1 score. Through the incorporation of signal indicators, ML models offer a robust approach to predicting changes in Bitcoin prices. The proposed ML approach achieves a test accuracy exceeding 65.15%, highlighting the potential synergy between traditional financial prediction tools and contemporary technology. This study underscores the potential effectiveness of combining ML models with technical indicators for forecasting Bitcoin price movements.

Machine learning is widely used in all industries. In order to speed up decision-making on the most likely course of action, machine learning have shown their effectiveness in analyzing perioperative effects. In several technological sectors, ML models have been utilized for a while to specify and rank dangerous threat variables. To handle forecasting problems, several different prediction approaches are regularly used. The work shows how ML models can forecast the frequency of COVID-19 instances in the future, which is now considered to be a serious threat to civilisation. In this study, we carried out a comparison analysis of two widely used machine learning models: SVM and linear regression. Each model projects three variables: the cumulative count of confirmed cases, the cumulative count of fatalities, and the cumulative count of recoveries over the next 30 days. The study's conclusions demonstrate that applying these tactics to the current COVID-19 pandemic scenario is a viable choice. To boost accuracy, we used two ML models. The results of the experiment indicate that when it comes to COVID-19 prediction, Linear Regression yields the best results, while SVM yields the worst.

Since the coronavirus pandemic started, many people have died due to the disease. The epidemic has been challenging to predict, as it progresses and spreads throughout the world. We used Auto-Regressive Integrated Moving Average (ARIMA) models to predict the outbreak of COVID-19 in the upcoming months in Morocco. In this work, we measured the effective reproduction number using the real data and the forecasted data produced by the two commonly used approaches, to reveal how effective the measures taken by the Moroccan government have been in controlling the COVID-19 outbreak. The prediction results for the next few months show a strong evolution in the number of confirmed and death cases in Morocco. We study the spread of COVID-19 in Morocco to see how many cases are discovered, recovered, and dead, and the forecasting of further cases is used as a basic novel method. It is based on time series models. We used coronavirus outbreak data from March 02, 2020, to August 04, 2021. ARIMA (Autoregressive integrated moving average) and Prophet time-series models are used to forecast the development of COVID-19, which is not a novel method. The mean absolute error, root mean square error, and coefficient of determination R2 were computed to assess the model's performance. Our study aims to provide a better understanding of the infectious disease outbreak that affected Morocco. It also provides information on the disease outbreak's epidemiology. Our study shows that the FBProphet model is more accurate in predicting the prevalence of COVID-19. It can help guide the government's efforts to prevent the virus' spread.

Mobile network operators (MNOs) continually look to improve their business and become more agile and competitive within the quickly developing telecommunications industry. In turn, telecommunication professionals put forward and adapt advanced models and frameworks to assist MNOs to achieve their goals. In this context, the Business Process Framework (eTOM) was established by the TeleManagement Forum and on which we base this work to suggest a mapping of this eTOM business process model upon the network functions virtualization (NFV) framework. Thereby, the main purpose is to design a hybrid architecture by the combination of the aforementioned frameworks for dynamic service delivery with improved resource performance and quality of service so as to fulfill some of the purposes of the 5 generation (5G) technology with regard to a telecommunication system managed and orchestrated in a virtualized environment. Indeed, MNOs will be in a position to scale mobile services up and down fast and reduce costs to better align them with network usage. These procedures are performed based on flexible service chaining through implemented NFV management and orchestration modules along with SDN controller functions. In this article, we outline possible designs and analyses of flexible mobile service chaining to support end-to-end network slicing for dynamic service provisioning in order to provide a novel approach in the framework of the 5G technology, more especially from data and signaling based network perspectives. We project our proposal onto the IP Multimedia Subsystem (IMS) core network to discuss both dynamic and static signaling service provisioning approaches. In this situation, we set up a testbed platform with the goal of assessing the behavior of a virtualized IMS system, in a static signaling service provisioning environment through two signaling service chains to validate part of our proposed approaches. Experiment results confirm that virtualized IMS signaling resource performance indicators could be enhanced in the case of unpredicted performance degradation of one of the virtualized signaling resources assigned to the IMS virtual network functions composing the main signaling service chain by instantiating a second signaling service chain. Finally, IMS application performance indicators are improved regarding registration delay and session setup time.

The 5th Generation (5G) network design is still under exploration and will require an innovative approach to network monitoring and management with a relentless focus on customer needs. Among the recent technologies, software-defined networking (SDN) and network functions virtualization (NFV), which are already offering network programmability and automation, have been proposed to improve resource management in 5G systems. Furthermore, the enhanced telecom operations map (eTOM) can be used to manage both End-to-End (E2E) service and end-customer experience. In this paper, we illustrate a novel SDN-based architecture for 5G networks that will be able to enhance quality-of-experience (QoE) monitoring and management by integrating SDN and eTOM. This architecture is deployed and evaluated by simulations of a service-level agreement (SLA) verification scenario. As a result, the implemented platform is capable of identifying SLA violation and can successfully enforce suitable mechanisms to dynamically adjust network parameters, thereby satisfying the SLA constraints.

In the latest years, service selection is becoming more and more important due to the significant effect of internet based services in the telecom industry. When it comes to selecting the best service, different candidate services with similar settings are proposed by different service providers. The selection should take into consideration the respect of the constraints of consumers in terms of Service Level Agreement contracts, what makes the modelling of the preferences of decision-makers for choice problems the main focus of this work. In order to model these preferences, we propose contextual preference functions based on machine learning techniques from neural networks. It will therefore be possible to further explain and decode preferences in order to facilitate negotiation and thus decision-making, thereby improving the quality of service providers while being on customer preferences.

New generation networks (NGN) are based on revolutionary aspects in IT and telecom fields. Huge throughput as well as a simplified infrastructure are based on the data center result of virtualization and cloud computing. Nevertheless, a very important question about the management part is always relevant, regardless of QoS management, of business, or of security. The chapter aims to propose a management architecture, which invokes the notion of broker and also of controller to centralize the management of networks and services.

With the on-going trends of the telecom services, the number of service providers with similar functionalities is undergoing a rapid growth. The customers face the difficulty to decide which service provider can satisfy their needs and full their requirements. Negotiating contracts between involved parts, and hiding heterogeneity in the distributed network environment has been challenging for telecom operators and service providers. Different languages exist to describe the Service Level Agreement (SLA), which is a contract between a service provider and a customer. However, since each service provider expresses his SLA in his own way, it disrupts the customer's choice of the best service provider, and leads to a bad contract management. In this respect, we propose a novel architecture for service selection, and SLA management between different stakeholders in our network architecture. The idea is to set up a smart broker where we implemented a Multi-Criteria Decision Making (MCDM) method to maximize utility function so that the customer can choose services with required QoS performances. We also came up with the idea of settling a negotiation model for the SLA, and a context based SLA contract ontology in IP Multimedia Subsystem (IMS) network is also proposed to provide users with a clear model to express their requirements and preferences. Moreover, we used the New Generation Operations Systems and Software (NGOSS) Framework to model and analyze networks and services actions. To better understand the relationship and the projection of NGOSS Framework and IMS platform, we introduce an SLA management and monitoring architecture in IMS network.

IMS is a standardized service architecture defined by 3GPP, ETSI, and IETF to provide multimedia services such as videoconferencing, VoD, and voice over IP. IMS is mainly based on the SIP protocol for session initialization. The convergence to full IP has advantages but also disadvantages. The latter are mainly inherited from the weaknesses of the IP protocol, in particular the QoS and the security aspects. It is in this context that this chapter is written. It has as main objective to analyze security in IMS networks as service layer in 4G to identify the most vulnerable points and propose security solutions that can be implemented without degrading the QoS.