Sadiq Alinsaif

The integration of Artificial Intelligence (AI) into Internet of Things (IoT) medical devices has revolutionized arrhythmia monitoring. However, the high computational and power demands of traditional Deep Learning (DL) models pose significant challenges for long-term, battery-operated smart electronics. Spiking Neural Networks (SNNs), inspired by the biological efficiency of the human brain, offer a promising solution. This paper reviews the intersection of SNNs, low-power IoT hardware, and biomedical signal processing. I examine the transition from frame-based to event-driven processing, and discuss the hardware–software co-design necessary for next-generation cardiac wearables.

Hybrid fiber-reinforced self-compacting concrete (HFR-SCC) exhibits exceptional mechanical performance and durability; however, accurately predicting its behavior after fire exposure remains challenging because of the highly nonlinear interactions among mixture constituents, fiber reinforcement, and elevated temperature. Existing machine learning studies primarily focus on individual mechanical properties and rarely integrate sustainability assessment with interpretable artificial intelligence tools suitable for engineering practice. This study presents an explainable hybrid machine learning framework for simultaneously predicting compressive strength, tensile strength, flexural strength, production cost, and embodied CO ₂ emissions of post-fire HFR-SCC. Five optimized hybrid models, namely ANN-PSO, RF-PSO, DT-PSO, XGB-PSO, and SVR-GB, were developed using a dataset of experimental samples with nine mixture-design variables. Model performance was comprehensively evaluated using MAE, RMSE, MAPE, SMAPE, MdAE, MBE, PBIAS, confidence interval analysis, composite ranking, and explainable artificial intelligence (AI) techniques including partial dependence analysis and feature importance evaluation. The ANN-PSO model achieved the highest accuracy for predicting compressive strength and production cost, while the RF-PSO model produced the most accurate CO ₂ emission estimates. Tensile and flexural strengths were also predicted with high reliability, as most hybrid models achieved coefficients of determination between 0.88 and 0.96. Sensitivity analysis demonstrated that temperature and cement content were the dominant parameters controlling post-fire mechanical performance, whereas optimized tree-based parameters substantially improved predictive performance. To facilitate practical implementation, a user-friendly graphical user interface (GUI) was developed for real-time prediction and mix evaluation. The proposed framework provides a reliable digital decision-support system that simultaneously optimizes structural materials performance, economic efficiency, and environmental sustainability, offering a practical pathway toward intelligent and low-carbon concrete design.

3D fiber-reinforced printed concrete (3D-FRPC) has emerged as a transformative construction technology that enables automated construction, reduces material waste, and facilitates complex geometries beyond the capabilities of conventional methods. However, accurately predicting its mechanical properties remains challenging due to the complex interactions among mixture parameters. This study develops hybrid machine learning models to predict the compressive strength (C.S) and flexural strength (F.S) of 3D-FRPC using a dataset of 464 samples collected from previous literature. Five hybrid models were implemented: CatBoost–WOA, LightGBM–HHO, Gradient Boosting–ALO, KNN–ABC, and DNN–DE. The results demonstrated that CatBoost–WOA achieved the highest prediction accuracy, outperforming the other models with R ² ≈ 0.99 and RMSE ≈ 2.9 MPa for compressive strength. It also produced the lowest prediction errors (MAE = 2.2 MPa, RMSE = 2.9 MPa, MAPE ≈ 7.2 %) and the best performance for flexural strength (MAE = 0.28 MPa and RMSE = 0.36 MPa). Overall model performance ranked as CatBoost–WOA > LightGBM–HHO > Gradient Boosting–ALO > DNN–DE > KNN–ABC. Furthermore, error distribution analysis showed that most prediction errors of CatBoost–WOA were concentrated within −5 to +5 MPa, indicating superior prediction stability. Sensitivity analysis revealed that the water-to-binder ratio (W/B) is the most influential parameter affecting compressive strength. Additionally, a dedicated graphical user interface (GUI) was developed to operationalize the optimized hybrid model, enabling rapid prediction of the mechanical properties of 3D fiber-reinforced printed concrete and transforming the proposed framework into a practical decision-support tool for intelligent mix design and performance optimization. The developed framework can support engineers in reducing experimental effort, accelerating mix design evaluation, and improving decision-making for practical 3D-FRPC applications.

Eggshell powder (ESP) and date palm ash (DPA) are increasingly used as sustainable cement substitutes in cementitious composites. This study used multi-expression programming (MEP) to develop prediction models due to its advantage of yielding model equations. The attributes of ESP and DPA-modified concrete chosen for modeling include compressive strength (C-S), eco-strength (E-C-S), and cost-strength ratio (C-S-R). Hyperparameters in MEP were fine-tuned to get the maximum accuracy for predictions. The models were validated using R 2 and statistical checks and analyzing the variance among predictions and real values. The MEP models were noted to be exact in estimating C-S, C-S-R, and E-C-S with an R 2 of 95, 93, and 92%, respectively, indicating good agreement with actual data. Additionally, the ±20% index analysis indicated that all values fall within the acceptable range, validating the model’s reliability. The mathematical expression-based MEP prediction models developed in this study can be applied to future C-S, C-S-R, and E-C-S predictions in ESP–DPA-modified concrete. These models are designed to operate with a predetermined set of input parameters and are incompatible with a variable set of inputs. Additionally, it is imperative to maintain consistency in the units of inputs to obtain precise predictions from the constructed models.

This study investigates techniques for medical image classification, specifically focusing on COVID-19 scans obtained through computer tomography (CT). Firstly, handcrafted methods based on feature engineering are explored due to their suitability for training traditional machine learning (TML) classifiers (e.g., Support Vector Machine (SVM)) when faced with limited medical image datasets. In this context, I comprehensively evaluate and compare 27 descriptor sets. More recently, deep learning (DL) models have successfully analyzed and classified natural and medical images. However, the scarcity of well-annotated medical images, particularly those related to COVID-19, presents challenges for training DL models from scratch. Consequently, I leverage deep features extracted from 12 pre-trained DL models for classification tasks. This work presents a comprehensive comparative analysis between TML and DL approaches in COVID-19 image classification.

Cardiac arrhythmias, characterized by deviations from the normal rhythmic contractions of the heart, pose a formidable diagnostic challenge. Early and accurate detection remains an integral component of effective diagnosis, informing critical decisions made by cardiologists. This review paper surveys diverse computational intelligence methodologies employed for arrhythmia analysis within the context of the widely utilized MIT-BIH dataset. The paucity of adequately annotated medical datasets significantly impedes advancements in various healthcare domains. Publicly accessible resources such as the MIT-BIH Arrhythmia Database serve as invaluable tools for evaluating and refining computer-assisted diagnosis (CAD) techniques specifically targeted toward arrhythmia detection. However, even this established dataset grapples with the challenge of class imbalance, further complicating its effective analysis. This review explores the current research landscape surrounding the application of graph-based approaches for both anomaly detection and classification within the MIT-BIH database. By analyzing diverse methodologies and their respective accuracies, this investigation aims to empower researchers and practitioners in the field of ECG signal analysis. The ultimate objective is to refine and optimize CAD algorithms, ultimately culminating in improved patient care outcomes.