Demyelinating diseases of the central nervous system (CNS), such as acute disseminated encephalomyelitis (ADEM), multiple sclerosis (MS), and neuromyelitis optica spectrum disorder (NMOSD), disrupt white matter integrity, leading to severe neurological impairments. Early and accurate detection is crucial but remains challenging due to subtle lesion characteristics, overlapping imaging features, and the time-consuming nature of manual segmentation. This study proposes a novel ensemble quantum–based deep learning (DL) framework that integrates Gaussian Gabor filtering for noise suppression and local texture enhancement, a Residual Feed-Forward Absolute Coordinate Vision Transformer (RFF-ACVit) for joint segmentation and feature extraction, and a hybrid Quantum Autoencoder–Quantum Convolutional Variational (QA–QCV) classifier for final diagnosis. The quantum modules leverage parallel quantum state evolution and variational inference to preserve critical spatial–contextual information while reducing dimensionality, thereby improving classification robustness. Interpretability is ensured through Shapley Additive Explanations (SHAP), enabling transparent identification of the most influential MRI features contributing to each prediction. On a real-world CHZU MRI dataset, the proposed model achieves 98.26% Dice score for segmentation and 99.35% classification accuracy, outperforming state-of-the-art CNN, Transformer, and conventional U-Net–based methods by a margin of 2–8%. This demonstrates its potential as an accurate, interpretable, and computationally efficient decision-support tool for CNS demyelinating disease detection.

This article aims to provide an innovative antenna suitable for ultrawideband (UWB) integrated with partial Ku band applications. The antenna is a two-port multiple-input multiple-output (MIMO) element that is printed on FR4 substrate and has a compact dimension of 39.5 × 22 mm2. An elliptically shaped defective ground structure forms the substrate's bottom plane, and a tapered feed connects to a distinctive floral slotted patch on top. The proposed 2-port MIMO is operating between 4.69 and 16.01 GHz, which makes it fit for UWB (3.1–10.6 GHz) and partly Ku band (12–18 GHz) applications. The antenna design incorporates a rectangular parasitic element among the radiating elements to achieve excellent diversity performance. These characteristics include an envelope correlation coefficient (ECC) of 0.0002, a diversity gain (DG) of approximately 10, and a mean effective gain (MEG) of less than − 3.01 dB. This makes the antenna an ideal choice for use in UWB integrated with a portion of Ku-band applications.

Breast cancer is a prevalent health issue among women, with one in eight women succumbing to the disease. A significant number of women neglect the necessity for breast cancer detection, as the treatment poses risks associated with exposure to radioactive radiation. Non-invasive procedures, hazardous radiation exposure, and limited diagnostic specificity for breast tumors hinder breast cancer screening methods. Currently, several medical technologies, including mammography, magnetic resonance imaging, computed tomography, positron emission tomography, and histopathological imaging, are being used to diagnose breast cancer early. Nevertheless, proficient radiologists or pathologists are necessary to interpret these imaging modalities. The process is challenging, costly, and prone to inaccuracies. Advances in machine learning and computing technology have transformed many facets of the world in the last ten years. Deep learning models have emerged as powerful tools in detecting tumors and predicting breast cancer, leveraging radiographic and histopathological images to achieve remarkable outcomes. Nevertheless, rigorous external validation is essential to ensure these advanced artificial intelligence technologies can be confidently integrated into clinical decision-making processes. The main objective of this research is to provide a critical analysis of findings on breast cancerous tumor classification using machine and deep learning algorithms. For this, we examined five prominent datasets: Wisconsin, SEER, ultrasound pictures, mammograms, and BreakHis histopathology images. The study reviewed the literature from the past decade (2015–2024) across multiple sources, including Springer, ScienceDirect, IEEE, PubMed, MDPI, Nature, Web of Science, Hindawi, and ArXiv. A detailed discussion was incorporated to elucidate ongoing research challenges and opportunities for future research in this burgeoning field.

This paper describes the design and construction of a small, flexible circular-slotted monopole antenna that works best at frequencies below 6 GHz in 5G New Radio (NR) and Wi-Fi 6 (IEEE 802.11ax) systems. The design includes a feed line that tapers in a triangular shape and a defected circular ground structure to improve radiation characteristics, optimize surface current distribution, and increase impedance bandwidth. The model is made on a polyimide substrate (18.5 × 27 × 0.6 mm³) that has an effective dielectric constant of 3.5 and a low loss tangent of 0.0027. This makes it lightweight, conformal, and low-profile, making it perfect for wearable and IoT integration. Simulated and observed data demonstrate that the impedance bandwidth ranges from 3.28 to 7.83 GHz, covering the whole sub-6 GHz 5G NR and Wi-Fi 6 frequency ranges. The maximum radiation efficiency is 99.2%, and the gain is 4.62 dBi. The 1-g Specific Absorption Rate (SAR) analysis conducted using a multilayer human phantom at different antenna–body separation distances revealed that the lowest SAR value is 0.0759 W/kg, occurs when the antenna is positioned 60 mm from the phantom. Mechanical deformation experiments with various bending radii demonstrate that |S11| remains stable, indicating that it can withstand real-world circumstances well. The findings reveal that the suggested antenna has a compact size, a large impedance bandwidth, and steady radiation characteristics. This makes it ideal for sub-6 GHz 5G, Wi-Fi 6, and wearable IoT applications. A time-domain investigation verifies the transient response and waveform quality in broadband and impulse-based communication situations.

Satellite-based video surveillance, sometimes known as “gazing,” is extremely useful for viewing, evaluating, and dynamically tracking developments on Earth. However, the tiny size and density of objects, overlapping targets, and unclear surroundings with variable illumination and complex backdrops make multi-object tracking in satellite movies particularly difficult. Limitations in bandwidth and computational capacity made accurate tracking and real-time processing increasingly challenging. Applications, including disaster response, traffic monitoring, defense, and security operations, depend on overcoming these obstacles. This work offers a novel framework to address these difficulties by merging sophisticated cross-frame connection techniques with frequency domain analysis using wavelet transform analysis. By separating high-frequency components and reducing noise, the wavelet transform improves the identification of small targets and makes it possible to recover fine-grained spatial and frequency data that are essential for reliable tracking. The system uses motion models and data association techniques to guarantee trajectory correctness and consistency, and it integrates cross-frame connections to create temporal continuity and preserve target identities over successive frames. The experimental results show significant increases in tracking performance, outperforming state-of-the-art methods in terms of multiple object tracking accuracy (MOTA), multiple object tracking precision (MOTP), and high tracking accuracy. These results demonstrate this proposed model's resilience and effectiveness in accurately identifying and following small targets in challenging satellite imagery settings. The accuracy achieved by the proposed method for the VISO, SATMTB, and Skysat-1 datasets is 96.82%, 95.26%, and 95.90%, respectively.

Detecting cardiac arrhythmia is now a very important field in computational biomedical engineering. This is because abnormal heart rhythms impact more than 59 million people throughout the world and traditional diagnostic methods have their own problems. Electrocardiography (ECG) and photoplethysmography (PPG) are still the main ways to monitor the heart in a clinical setting or on a wearable device. However, rapid advancements in signal processing, machine learning (ML), deep learning (DL), and the new field of quantum machine learning (QML) have made arrhythmia diagnosis a computationally accelerated, algorithm-driven process. This study gives a comprehensive overview of the latest developments in AI-based arrhythmia detection, with a focus on how these methods improve early diagnosis, reliability, and understanding in cardiac monitoring systems. The work rigorously analyzes publicly accessible ECG and PPG datasets, feature extraction techniques, classification methods, and performance evaluation criteria utilized in the literature. The emphasis is on the amalgamation of Quantum Machine Learning (QML) with deep neural architectures, where quantum computation enhances feature representation, dimensionality reduction, and classification accuracy via quantum parallelism and entanglement. The survey includes studies published in well-known databases like IEEE, Springer, ScienceDirect, PubMed, Frontiers, and MDPI between 2014 and 2025. The studies used keywords like “cardiac arrhythmia detection,” “atrial fibrillation,” “irregular heartbeat diagnosis,” “machine learning,” “deep learning,” “quantum machine learning,” “quantum hybrid neural networks,” and “hyperparameter optimization techniques.” This study brings together the latest findings, points out areas where more research is needed, and suggests where AI-driven cardiac diagnostics could go in the future. It ends by suggesting a mixed framework that combines ML, DL, and QML methods to make arrhythmia prediction and classification more accurate, scalable, and reliable.

As childhood vaccination is vital for children to prevent them from vaccine preventable diseases, vaccine hesitancy (VH) is a phenomenon that can jeopardize this preventive mechanism. This study aims to develop an instrument to predict VH among parents towards childhood immunization by using machine learning (ML) algorithms. In this study, the approach of predicting VH was to focus on attitude, behavior and practice through the administration of a questionnaire which was verified by statistical analysis and ML algorithms. The researchers developed a 26-item instrument adapted from two other studies. Experts from three different fields reviewed the instrument for content validity. From the pilot study, a 13-item instrument was generated and has a Cronbach alpha value of 0.850 for reliability. The instrument was applied to 510 respondents who are parents attending the Obstetrics and Gynecology and Pediatric Clinics of the state referral hospital, and have children between the ages of 0 and 15 years old. The data collected was subjected to 10 ML algorithms. It was found that in terms of accuracy, the logistic regression with bagging method produced the best results with 99.02% for the hold-out method and 97.45% for the 10-fold cross-validation method. The results of our study show that there is potential of the instrument to anticipate parental VH in the local situation. The instrument’s prospects can be further enhanced if its performance is validated against an objective parameter such as vaccination records.

Path planning is a non-deterministic polynomial-time rigid problem. This research compares three distinct trajectory control and path planning algorithms: the spar-row search algorithm (SSA), the hybrid ant colony optimization and genetic algorithm (ACO-GA), and the ant colony optimization (ACO) method. We talk about the best algorithm for a static or dynamic environment. Gaining further insight into how metaheuristic algorithms work when resolving shortest path problems is the aim of this study. In order to tabulate and discuss the results, the convergence curve is plotted and a pixmap is created. The results showed that the SSA had a path time that was 0.07 s faster than the ACO and 0.58 s faster than the ACO-GA. The length of the SSA-generated trajectory optimization algorithm is the overall shortest and smoothest. Moreover, SSA had the maximum path value, the lowest path time, and the fewest iterations—less than 35. For SSA, the angle of rotation worked best since it could determine the destination with the greatest amount of efficiency. Therefore, SSA algorithm yields better results compared to the other two algorithms.

This work presents the design of a flexible, low-profile radiator for wearable applications. The ability to put the antenna on the body is an essential requirement for on-body applications. Jeans material with 0.6 mm thickness, the tangent loss is 0.02, and is 1.6 is used to design with a size of 23×25×0.6 mm3 . A fan shaped patch is integrated on the substrate with a defected ground structure (DGS) is printed on back of substrate, which is used to achieve a bandwidth from 3.4-4.2 GHz for |s11|<-10 dB with an efficiency 96%, operated for 3.3-3.6 GHz (WiMAX), and 3.7-4.2 GHz (C-band) applications.

In this study, the design and analysis of a compact conformal antenna for use in drone applications are presented. With a total dimension of 35 × 52 mm2, the antenna had a substrate dielectric constant of 2.2 and a thickness of 1.6 mm. Additionally, the antenna model had a truncated square patch in addition to two slots in the shape of an I. Software that is part of the Computer Simulation Technology (CST) industry is used to simulate the antenna model. The antenna is capable of achieving broad band performance, which allows it to successfully cover the frequency range of 5.2-5.4 GHz in the common band. The antenna that was designed displays outstanding return loss characteristics of -19.0dB at the corresponding frequency range, as determined by analysis and simulation. This demonstrates that the impedance matching is strong and that the power transmission is efficient. Because of its small size and low weight, the microstrip antenna is the ideal candidate for incorporation into a wide variety of communication and drone systems.

This paper presents the design and implementation of an IoT-based system for home monitoring and appliance control, aimed at enhancing security, energy efficiency, and user convenience. The system combines intelligent sensors, actuators, and a cloud-based dashboard to support real-time monitoring and remote control via web and mobile applications. A wireless communication network ensures seamless device interaction and secure data exchange. Experimental results demonstrate the system's effectiveness in automatically controlling appliances and generating timely alerts in response to abnormal conditions, such as gas leaks or intrusions. The proposed solution offers a practical and scalable approach to home automation, contributing to the development of intelligent residential environments.

Electrocardiogram (ECG) analysis is crucial for the identification and classification of cardiac arrhythmias; it is a gold standard for cardiovascular mortality. In this work, a well-structured classification process for identifying cardiac arrhythmias using the MIT-BIH dataset. This dataset consists of a raw ECG signal pre-processed with wavelet transform to remove noise without affecting diagnostic features. The PanTompkins algorithm identified R-peaks of the segmented ECG signal. Annotations of heartbeat are divided into five classes: normal, ventricular ectopy, supraventricular ectopy, fusion, and unclassified beats, according to the Association for the Advancement of Medical Instrumentation (AAMI EC57) standards. The imbalance in classes is balanced using a hybrid method of Synthetic Minority Over-sampling Technique (SMOTE) and random undersampling. The balanced classes were trained and tested using machine learning models to perform the detection of arrhythmias that enhance the accuracy, precision, and recall. The Random Forest classifier is an ensemble method performs the classification of arrhythmia with an accuracy of 99.56%.

Cardiovascular diseases claim approximately 17.9 million lives annually, with heart attacks and strokes accounting for over 80% of these deaths. Key risk factors, including hypertension, hyperglycemia, dyslipidemia, and obesity, are identifiable, offering opportunities for timely intervention and reduced mortality. Early detection of heart disease enables individuals to adopt lifestyle changes or seek medical treatment. However, conventional diagnostic methods, such as electrocardiograms—commonly used in clinics and hospitals to detect abnormal heart rhythms—are not effective in identifying actual heart attacks. Additionally, angiography, while more precise, is an invasive method, financial strain on patients, and high chances of incorrect diagnosis, highlighting the need for alternative approaches. The main goal of this study was to assess the accuracy of machine learning techniques, including both individual and combined classifiers, in early detection of heart diseases. Furthermore, the study aims to highlight areas where additional research is necessary. Our investigation covers a decade period from 2014 to 2024, including a thorough review of pertinent literature from international conferences and top journals from the databases like Springer, ScienceDirect, IEEEXplore, Web of Science, PubMed, MDPI, Hindawi and so on. The following keywords were used to search the articles: heart disease risk, heart disease prediction, data mining, data preprocessing, machine learning algorithms, ensemble classifiers, deep learning algorithms, feature selection, hyperparameter optimization techniques. We examine the methodologies used and evaluate their effectiveness in predicting cardiovascular conditions. Our findings reveal notable progress in applying machine learning and deep learning in cardiology. The study concludes by proposing a framework that incorporates current machine learning techniques to enhance heart disease prediction.

Joint Source-Channel Coding (JSCC) has emerged as a major trend with the introduction of advanced technologies. High reconstruction quality, low cliff effect, and high spectral efficiency are the main goals behind the development of these techniques. However, traditional approaches have a large computation overhead and decoding errors. A novel Double Deep Polar Coding/DECoding (DDP-Codec) for JSCC is presented in this paper. The proposed DDP-Codec encoder structure consists of two cascaded polar codes. Deep polar code is an advanced version of polar codes in which the performance is enhanced with additional layers of encoding /decoding processes. Deep polar codes are used for channel error correction, in which the polar code compresses the source information. For source and channel decoding, the proposed model presents a joint Turbo-Like Successive Cancelation List with Back Propagation Parity Checks (TL-SCL-BPPC) decoder. Internal iterations are carried out independently by the source and channel decoders. The exchange of soft information between joint decoders is performed in an external iteration. In order to obtain the original information, the soft information of the channel polar code generated by the source decoding process using the external iteration of the TL-SCL-BPPC. The performance of the proposed approach is evaluated with various performance metrics and compared with state-of-the-art approaches. By using the proposed approach, the Frame Error Rate (FER), Mean Square Error (MSE), BiLingual Evaluation Understudy (BLEU), and average bits are improved by 0.05%, 0.10%, 0.09%, and 25%, respectively.

It is difficult for automated driving systems, or advanced driver assistance systems, to recognize and comprehend their surroundings. This paper proposes a transformer model-based approach for road object recognition using sensor fusion. Initially, data from the camera and millimeter-wave (mmWave) radar are simultaneously acquired and pre-processed. Since direct point cloud-to-image fusion is difficult for fusion object detection models, the radar point clouds are then circularly projected onto a 2-dimensional (2D) plane. Then, both the camera image and radar projection image enter different branches of the feature extraction model, utilizing a dual-path vision transformer (DualP-ViT) to complete feature extraction and fusion. The items are recognized after going through several layers of encoders and decoders. An encoder decoder-based vision transformer (EDViT) provides accurate measures of distance and velocity. Also, the vision sensors (cameras) produce high-resolution images with rich visual information. The proposed approach is implemented on the nuScenes dataset, and the performance is evaluated based on object detection metrics. The mean Average Precision (mAP), NuScenes Detection Score (NDS), Planning KL-Divergence (PKL), accuracy, precision, recall, f1-score, and latency performance obtained with the proposed approach is 59, 68, 0.6, 80, 79, 80, 78.9, and 10 ms. In the proposed approach, the robustness and accuracy of object detection is improved.

Stroke is recognized as a source of numerous impairments, encompassing deficits in physical, motor, and emotional functions in affected individuals. While the visible manifestations of a stroke are evident, the internal effects on the brain remain mostly enigmatic. Research has shown that utilizing motor imagery tasks via Electroencephalogram (EEG) bio signals achieves a 10% increase in accuracy relative to traditional techniques. This research work aims to employ feature extraction techniques on motor imaging tasks combining right-and left-hand grasping, utilizing motor imagery-based EEG data to extract the most pertinent features from two distinct datasets. One dataset comprises individuals with stroke, and the other consists of healthy individuals. Techniques such as the Common Spatial Filter (CSP) and the Filter Bank Common Spatial Filter (FBCSP) are employed to extract relevant features from the processed and filtered data. Three supervised machine learning algorithms, including Support Vector Machines (SVM), Linear Discriminant Analysis (LDA), and Gaussian Naïve Bayes (GNB), have been employed for data classification. A comparative study has been conducted to understand the fundamental differences in the EEG signals between stroke patients and healthy individuals. The findings indicated that the FBCSP approach surpassed CSP in both categories of patients, with the SVM achieving an accuracy of up to 98.86% in classifying motor imagery tasks. This comparative study enhances our understanding of Brain-Computer Interface (BCI) systems and motor rehabilitation methods by elucidating critical differences between EEG data from stroke patients and healthy individuals.

Due to the abundance and richness of spectral-spatial information, hyperspectral images (HSIs) obtained from hyperspectral imaging have been widely used in a variety of applications, including target or anomaly identification. However, due to its low processing complexity, onboard real-time anomaly identification has always been challenging in hyperspectral image analysis. To achieve high detection accuracy, most existing anomaly detection systems inevitably compromise on high computational complexity. In this paper, a new lightweight field-programmable gate array (FPGA) accelerator is proposed for hyperspectral anomaly detection using HSIs. The proposed approach consists of two stages. In the first stage, average fusion is used to reduce the dimensions of the HSIs. In the second stage, an optimized TinyYOLOv3 accelerator is utilized to extract features and detect anomalies. This optimized TinyYOLOv3 accelerator uses a hardware-friendly shift-based floating-fixed multiply accumulator (MAC) operator and a shift-based quantization method. The shift-based floating-fixed MAC operator is further optimized using a compact LUT-based multiplier (C-LUT-MUL) and an effective floating point adder. The proposed lightweight FPGA Accelerator is implemented on the coding tool Xilinx Verilog using San Diego, Urban-Beach, and EI Segundo datasets. The evaluation results reveal that the proposed accelerator has a higher resource consumption and processing speed (62.5 FPS) while maintaining maximum detection accuracy. This shows the benefits of the proposed lightweight FPGA accelerator over existing research.

Sudden cardiac arrest (SCA) is a serious heart problem that occurs without symptoms or warning. SCA causes high mortality. Therefore, it is important to estimate the incidence of SCA. Current methods for predicting ventricular fibrillation (VF) episodes require monitoring patients over time, resulting in no complications. New technologies, especially machine learning, are gaining popularity due to the benefits they provide. However, most existing systems rely on manual processes, which can lead to inefficiencies in disseminating patient information. On the other hand, existing deep learning methods rely on large data sets that are not publicly available. In this study, we propose a deep learning method based on one-dimensional convolutional neural networks to learn to use discrete fourier transform (DFT) features in raw electrocardiogram (ECG) signals. The results showed that our method was able to accurately predict the onset of SCA with an accuracy of 96% approximately 90 minutes before it occurred. Predictions can save many lives. That is, optimized deep learning models can outperform manual models in analyzing long-term signals.

Diabetes Mellitus is a prevalent condition globally, marked by elevated blood sugar levels resulting from either insufficient production of insulin or the body cells' inability to respond appropriately to released insulin. For people with diabetes to lead healthy, normal lives, early identification and treatment of the condition are essential. With the need to move away from current traditional procedures, towards a noninvasive methodology, machine learning and data mining technologies can be very useful in the classification of diabetes. Creating an effective machine learning model for the classification of diabetes mellitus was the primary goal of this research. This work is primarily carried out on combined Pima Indian diabetes dataset and German Frankfurt diabetes dataset. The class imbalance issue has been resolved using Synthetic Minority Oversampling Technique. One-hot encoding is applied to convert categorial features to numerical and various single and ensemble classifiers with the best hyperparameters obtained using GridSearchCV method were employed on the pre-processed dataset. With an AUC of 0.98 and maximum accuracy of 98.79%, the Random Forest ensemble technique outperformed the other models, according to the experimental results. As a result, the algorithm might be used to predict diabetes and alert doctors to serious cases that call for emergency care.

Medical science-related studies have reinforced that the prevalence of coronary heart disease which is associated with the heart and blood vessels has been the most significant cause of health loss and death globally. Recently, data mining and machine learning have been used to detect diseases based on the unique characteristics of a person. However, these techniques have often posed challenges due to the complexity in understanding the objective of the datasets, the existence of too many factors to analyze as well as lack of performance accuracy. This research work is of two-fold effort: firstly, feature extraction and selection. This entails extraction of the principal components, and consequently, the Correlation-based Feature Selection (CFS) method was applied to select the finest principal components of the combined (Cleveland and Statlog) heart dataset. Secondly, by applying datasets to three single and three ensemble classifiers, the best hyperparameters that reflect the pre-eminent predictive outcomes were investigated. The experimental result reveals that hyperparameter optimization has improved the accuracy of all the models. In the comparative studies, the proposed work outperformed related works with an accuracy of 97.91%, and an AUC of 0.996 by employing six optimal principal components selected from the CFS method and optimizing parameters of the Rotation Forest ensemble classifier.

A collaborative robot, or cobot, enables users to work closely with it through direct communication without the use of traditional barricades. Cobots eliminate the gap that has historically existed between industrial robots and humans while they work within fences. Cobots can be used for a variety of tasks, from communication robots in public areas and logistic or supply chain robots that move materials inside a building, to articulated or industrial robots that assist in automating tasks which are not ergonomically sound, such as assisting individuals in carrying large parts, or assembly lines. Human faith in collaboration has increased through human–robot collaboration applications built with dependability and safety in mind, which also enhances employee performance and working circumstances. Artificial intelligence and cobots are becoming more accessible due to advanced technology and new processor generations. Cobots are now being changed from science fiction to science through machine learning. They can quickly respond to change, decrease expenses, and enhance user experience. In order to identify the existing and potential expanding role of artificial intelligence in cobots for industrial applications, this paper provides a systematic literature review of the latest research publications between 2018 and 2022. It concludes by discussing various difficulties in current industrial collaborative robots and provides direction for future research.

Heart disease is the leading cause of death and killing millions of people every year around the world. Various automated intelligent systems to predict the heart disease risk have been developed by current research works. However, these studies have drawbacks such as the inability to pick important features, lack of hyperparameter optimization, and varied performance from one model to another. In this work, proposed an unconventional feature engineering in which a Principal Component Analysis was performed on heart dataset to extract the transformed features and selected significant ones among them using Relief method. The hyperparameters of Support Vector Machine, K-Nearest Neighbors, J4S Decision Tree, AdaBoost Ml, Bagging, and Rotation Forest classifiers were optimized and performed machine learning classification using 10-fold cross-validation. The proposed work produced highest accuracy of 98.43% and AUC of 0.996 using KNN, while the Rotation Forest reached the accuracy of 98.25% and best AUC of 0.997.

Cardiovascular diseases (CVDs) kill about 20.5 million people every year. Early prediction can help people to change their lifestyles and to ensure proper medical treatment if necessary. In this research, ten machine learning (ML) classifiers from different categories, such as Bayes, functions, lazy, meta, rules, and trees, were trained for efficient heart disease risk prediction using the full set of attributes of the Cleveland heart dataset and the optimal attribute sets obtained from three attribute evaluators. The performance of the algorithms was appraised using a 10-fold cross-validation testing option. Finally, we performed tuning of the hyperparameter number of nearest neighbors, namely, ‘k’ in the instance-based (IBk) classifier. The sequential minimal optimization (SMO) achieved an accuracy of 85.148% using the full set of attributes and 86.468% was the highest accuracy value using the optimal attribute set obtained from the chi-squared attribute evaluator. Meanwhile, the meta classifier bagging with logistic regression (LR) provided the highest ROC area of 0.91 using both the full and optimal attribute sets obtained from the ReliefF attribute evaluator. Overall, the SMO classifier stood as the best prediction method compared to other techniques, and IBk achieved an 8.25% accuracy improvement by tuning the hyperparameter ‘k’ to 9 with the chi-squared attribute set.

Cardiovascular diseases (CVDs) are killing about 17.9 million people every year. Early prediction can help people to change their lifestyles and to endure proper medical treatment if necessary. The data available in the healthcare sector is very useful to predict whether a patient will have a disease or not in the future. In this research, several machine learning algorithms such as Decision Tree (DT), Discriminant Analysis (DA), Logistic Regression (LR), Naïve Bayes (NB), Support Vector Machines (SVM), K-Nearest Neighbors (KNN), and Ensemble were trained on Cleveland heart disease dataset. The performance of the algorithms was evaluated using 10-fold cross-validation without and with Principal Component Analysis (PCA). LR provided the highest accuracy of 85.8% with PCA by keeping 9 components and Ensemble classifiers and attained an accuracy of 83.8% using a Bagged tree with PCA by keeping 10 components.

Heart disease risk prediction is very important as it is one of the primary causes of sudden death in the world. Early-stage prediction can save the lives by undergoing appropriate diagnosis steps or making necessary changes in their lifestyles. Recent studies have focused on the use of data mining and machine learning in the detection of diseases based on specific features of a person. The Rotation Forest, a tree-based ensemble classifier that uses Principal Component Analysis for feature extraction, is proposed to improve the prediction accuracy of heart disease risk. The Statlog heart dataset has been selected from the publicly available UCI machine learning repository in this research work. The dataset was trained with a Rotation Forest ensemble classifier with default base classifier J48, and then, Random Forest on full features and selected features obtained from One Rule and Support Vector Machines attribute evaluators. The performance of the Rotation Forest was compared with the standard machine learning classifiers, Naïve Bayes, Logistic Regression, Support Vector Machines, K-Nearest Neighbors, AdaBoostM1, and Bagging. The Rotation Forest algorithm with Random Forest provided the highest accuracy of 94.44% and area under the ROC curve 0.980 on selected features of the Statlog dataset from the One Rule method.

Heart disease, one type of cardiovascular illness, is the leading cause of mortality for many individuals around the world. Early prediction of heart disease can help people to endure appropriate medical treatment and to save lives. Recent studies have focused on the use of data mining and machine learning in the detection of diseases based on specific features of a person. In this research, prepared an integrated heart dataset of 1190 observations from the Cleveland, Hungarian, Long Beach VA, Switzerland, and Statlog. Numerous machine learning classifiers, Decision Tree, Discriminant Analysis, Logistic Regression, Naïve Bayes, Support Vector Machines, k-Nearest Neighbors, Bagged Trees, Optimizable Tree, and Optimizable k-Nearest Neighbors are trained using 10-fold cross-validation for efficient heart disease risk prediction on the Correlation-based Feature Selection optimal set of the integrated heart dataset. Finally, performed a comparative analysis with and without feature selection. The Optimizable k-Nearest Neighbors algorithm achieved an utmost accuracy of 95.04%, area under the ROC curve of 0.99 on the Correlation-based Feature Selection optimal set, and that of 90.34%, 0.96 respectively, on full features.