Search

Heart disease is one of the most widespread global health issues, it is the reason behind around 32 % of deaths worldwide every year. The early prediction and diagnosis of heart diseases are critical for effective treatment and sickness management. Despite the efforts of healthcare professionals, cardiovascular surgeons and cardiologists' misdiagnosis and misinterpretation of test results may happen every day. This study addresses the growing global health challenge raised by Cardiovascular Diseases (CVDs), which account for 32 % of all deaths worldwide, according to the World Health Organization (WHO). With the progress of Machine Learning (ML) and Deep Learning (DL) techniques as part of Artificial Intelligence (AI), these technologies have become crucial for predicting and diagnosing CVDs. This research aims to develop an ML system for the early prediction of cardiovascular diseases by choosing one of the powerful existing ML algorithms after a deep comparative analysis of several. To achieve this work, the Cleveland and Statlog heart datasets from international platforms are used in this study to evaluate and validate the system's performance. The Cleveland dataset is categorized and used to train various ML algorithms, including decision tree, random forest, support vector machine, logistic regression, adaptive boosting, and K-nearest neighbors. The performance of each algorithm is assessed based on accuracy, precision, recall, F1 score, and the Area Under the Curve metrics. Hyperparameter tuning approaches have been employed to find the best hyperparameters that reflect the optimal performance of the used algorithms based on different evaluation approaches including 10-fold cross-validation with a 95 % confidence interval. The study's findings highlight the potential of ML in improving the early prediction and diagnosis of cardiovascular diseases. By comparing and analyzing the performance of the applied algorithms on both the Cleveland and Statlog heart datasets, this research contributes to the advancement of ML techniques in the medical field. The developed ML system offers a valuable tool for healthcare professionals in the early prediction and diagnosis of cardiovascular diseases, with implications for the prediction and diagnosis of other diseases as well.

Convolution Neural Network (CNN) models have gained ground in research activities particularly in medical images used for Diabetes Retinopathy (DR) detection. X-ray, MRI, and CT scans have all been used to validate CNN models, with classification accuracy generally reaching that of trained doctors. It is mandatory to evaluate the strength of CNN models used in medical tasks against adversarial attacks especially in healthcare, that is to say, the security of such models is becoming extremely relevant to the diagnosis as this latter will guide high-stakes decision-making. However, little study has been conducted to better comprehend this issue. This paper focuses on MobileNet CNN architecture in order to investigate its vulnerability against fast gradient sign methods (FGSM) adversarial attacks. For this end, a Neural Structure Learning (NSL) and a Multi-Head Attention (MHA) have been used to effectively reduce the vulnerability against attack by end-to-end CNN training with adversarial neighbors that produce adversarial perturbations on optical coherence tomography (OCT) images. With suggested model NSL-MHA-CNN, there has been an ability to maintain model performance on adversarial attack without increasing cost of training. Through theoretical assistance and empirical validation, it was possible to examine the stability of MobileNet architecture and demonstrate its susceptibility, particularly to adversarial attack. The experiments in this paper show that indiscernible degrees of perturbation $\varepsilon < 0.01$ were sufficient to cause a task failure resulting to misclassification in majority of the time. Moreover, empirical simulation shows that the proposed approach advanced in this paper can be an effective method to defense against adversarial attack at level of CNN model testing.

In the context of COVID-19 pandemic, the Moroccan Interior and Health Ministries have proposed to use the health pass with a QR code to identify vaccinated people. Additionally, the government suggested a mobile application to control the health passport authenticity. However, the key problem is the possibility of anyone scanning the QR code and figuring out citizens' private information, causing severe issues about individual privacy. In this work, the main contribution is integrating a private Blockchain-based digital health passport to ensure high protection of sensitive information, security and privacy among all the actors (Government, Ministry of Interior, Ministry of Health, verifiers) that comply with the CNDP (National Commission for the Control of Personal Data Protection) and the Moroccan Law 09–08. In our proposed architectural framework solution, we identify two types of actors: authorized and unauthorized, to limit and control access to the citizens' personal information. Besides, to preserve individuals' privacy, we adopt on-chain and off-chain storage (Interplanetary File Systems IPFS). In our case, smart contracts improve security and privacy in the health passport verification process. Our system implementation describes the proposed solution to grant individual privacy. To verify and validate our approach, we used Remix-IDE and Ethereum Blockchain to build smart contracts.

With the advancements in technology and the growing demand for online education, Virtual Learning Environments (VLEs) have experienced rapid development in recent years. This demand was especially evident during the COVID-19 pandemic. The incorporation of new technologies in VLEs provides new opportunities to better understand the behaviors of learners. Identifying the learning styles (LSs) of learners can greatly impact the learning process and enhance the effectiveness and satisfaction of both learners and teachers in Massive Online Open Courses (MOOCs). In this study, an approach to automatically recognize the LSs of learners based on the amount of data interactions generated in MOOC is presented. The Felder Silverman Learning Style Model (FSLSM) is used as the basis for prediction, as it is one of the most widely used models in VLEs. The data collected from the learning activities in the XuetanX platform from 08-2016 to 08-2017 was prepared using the Get and Structure Data (GSD) algorithm and then clustered using the K-means algorithm based on the Felder & Silverman Model. The learner's degree performance in each learning style was analyzed using the confusion matrix, learning curves, and performance metrics (accuracy, precision, recall, and macro/micro-averaged precision) for the neural network (NN), decision tree (DT), Random Forests (RF), Naive Bayes (NB), and K-nearest neighbors (KNN) algorithms. The evaluation results showed that the DT achieved a high accuracy rate of over 99% in predicting learners' learning styles.

The current healthcare systems are facing many issues in terms of data management, data sharing, information security and patient privacy, data immutability, trust, and transparency. In addition, the multiple existing healthcare systems are centralized which complicates the healthcare professionals, patients in managing their data and causes several problems. Blockchain technology as a decentralized peer-to-peer network has the power to digitalize and transform the manner that the data are managed in the healthcare industry, in this regard, is one such domain that might benefit from Blockchain technology in different manners. This paper aims to improve a review of recent works on Blockchain-based healthcare applications.

Atherosclerosis diagnosis is an indistinct and complex cognitive process. Artificial intelligence methods, such as machine learning algorithms, have proven their efficiency in Medical Diagnosis Support Systems (MDSS). In this paper, we developed a novel machine learning MDSS to boost the diagnosis of cardiovascular diseases. Our study performed using 835 patient medical records that suffer from atherosclerosis, usually caused by coronary artery diseases (CAD), collected from three databases. The system input layer includes several input variables based on three databases, the Cleveland heart disease, Hungarian, and Z-Alizadeh Sani databases. Seven independent classification methods are applied to assess the system: Artificial Neural Network (ANN), K-Nearest Neighbor (KNN), Support Vector Machine (SVM), Decision Tree (DT), Naïve Bayes (NB), Classification Ensemble (CE), and Discriminant Analysis (DA) algorithms. The robustness of the proposed methods was evaluated through several performance measures. The results showed that the proposed MDSS reached an accuracy of (98%), which is a higher accuracy than the existing approaches. These results are a promising step toward facilitating large-scale clinical diagnostics for atherosclerosis diseases.