Background/Objectives: Artificial intelligence-based software as a medical device (AI-SaMD) refers to AI-powered software used for medical purposes without being embedded in physical devices. Despite increasing approvals over the past decade, research in this domain—spanning technology, healthcare, and national security—remains limited. This research aims to bridge the existing research gap in AI-SaMD by systematically reviewing the literature from the past decade, with the aim of classifying key findings, identifying critical challenges, and synthesizing insights related to technological, clinical, and regulatory aspects of AI-SaMD. Methods: A systematic literature review based on the PRISMA framework was performed to select the relevant AI-SaMD studies published between 2015 and 2024 in order to uncover key themes such as publication venues, geographical trends, key challenges, and research gaps. Results: Most studies focus on specialized clinical settings like radiology and ophthalmology rather than general clinical practice. Key challenges to implement AI-SaMD include regulatory issues (e.g., regulatory frameworks), AI malpractice (e.g., explainability and expert oversight), and data governance (e.g., privacy and data sharing). Existing research emphasizes the importance of (1) addressing the regulatory problems through the specific duties of regulatory authorities, (2) interdisciplinary collaboration, (3) clinician training, (4) the seamless integration of AI-SaMD with healthcare software systems (e.g., electronic health records), and (5) the rigorous validation of AI-SaMD models to ensure effective implementation. Conclusions: This study offers valuable insights for diverse stakeholders, emphasizing the need to move beyond theoretical analyses and prioritize practical, experimental research to advance the real-world application of AI-SaMDs. This study concludes by outlining future research directions and emphasizing the limitations of the predominantly theoretical approaches currently available.

The article reflects on the classification of brain tumors where several deep learning (DL) approaches are used. Both primary and secondary brain tumors reduce the patient’s quality of life, and therefore, any sign of the tumor should be treated immediately for adequate response and survival rates. DL, especially in the diagnosis of brain tumors using MRI and CT scans, has applied its abilities to identify excellent patterns. The proposed ensemble framework begins with the image preprocessing of the brain MRI to enhance the quality of images. These images are then utilized to train seven DL models and all of these models recognize the features related to the tumor. There are four models which are General, Glioma, Meningioma, and Pituitary tumors or No Tumor model, which helps in reaching a joint profitable prediction and concentrating solely on the strength of the estimation and outcome. This is a significant improvement over all the individual models, attaining a 99. 43% accuracy. The data used in this research was gotten from Kaggle website and comprised of 7023 images belonging to four classes. The further works will be to increase the data size more investigate other DL architectures, further enhance real time detection to enhance on the diagnostic scans’ accuracy and overall relevancy to clinical practice.

While many articles have analyzed the effectiveness of the policies that aimed to limit the spread of COVID-19, very little research work has examined the determinants that drove these policies. Therefore, we proposed to study the determinants that led government authorities to implement more or less restrictive policies to limit the spread of the pandemic. Using the COVID-19 stringency index, we highlighted a positive effect of the incidence rate on the stringency level. Patient capacity in intensive care units was also a key variable. This is indicative of the capacity of countries to have a sufficient and appropriate health system to absorb such pandemic crises. On the other hand, we show that epidemiological data regarding the risk of excess mortality (diabetes, cancer, and cardiovascular pathologies) had a negative effect. We conclude by recalling the importance of policy coordination between countries when it comes to lowering the stringency levels of measures, in order to avoid a resurgence of the epidemic.

Stroke is a main risk to life and fitness in current society, particularly in the aging population. Also, the stroke is recognized as a cerebrovascular accident. It contains a nervous illness, which can result from haemorrhage or ischemia of the brain veins, and regular mains to assorted motor and cognitive damages that cooperate with functionality. Screening for stroke comprises physical examination, history taking, and valuation of risk features like age or certain cardiovascular illnesses. Symptoms and signs of stroke include facial weakness. Even though computed tomography (CT) and magnetic resonance imaging (MRI) are standard diagnosis techniques, artificial intelligence (AI) systems have been constructed based on these methods, which deliver fast detection. AI is gaining high attention and is being combined into numerous areas with medicine to enhance the accuracy of analysis and the quality of patient care. This paper proposes an enhancing neurological disease diagnostics fusion of transfer learning for acute brain stroke prediction using facial images (ENDDFTL-ABSPFI) method. The proposed ENDDFTL-ABSPFI method aims to enhance brain stroke detection and classification models using facial imaging. Initially, the image pre-processing stage applies the fuzzy-based median filter (FMF) model to eliminate the noise in input image data. Furthermore, fusion models such as Inception-V3 and EfficientNet-B0 perform the feature extraction. Moreover, the hybrid of convolutional neural network and bidirectional long short-term memory (CNN-BiLSTM) model is employed for the brain stroke classification process. Finally, the multi‐objective sailfish optimization (MOSFO)-based hyperparameter selection process is carried out to optimize the classification outcomes of the CNN-BiLSTM model. The simulation validation of the ENDDFTL-ABSPFI technique is investigated under the Kaggle dataset concerning various measures. The comparative evaluation of the ENDDFTL-ABSPFI technique portrayed a superior accuracy value of 98.60% over existing methods.

Internet of Vehicles (IoV) is a multi-node network which switches data in an open and wireless environment. Numerous interaction activities occur among IoV entities to share significant information, which is essential for network operation. As part of intellectual transportation, IoV is a hot topic for researchers because it faces numerous unresolved challenges, particularly regarding privacy and security. The development of recent malicious software with the expanding use of digital services has increased the likelihood of stealing data, corrupting data, or other cybercrimes by malware threats. Hence, malicious software should be perceived previously. It impacts a vast amount of computers. Researchers have proposed numerous malware detection solutions for the past few years. Machine learning (ML) and deep learning (DL)-based detection models can decrease analysis time and increase malware detection accuracy. This study proposes a novel Malware Detection Model in the Internet of Vehicles Using Deep Learning-Based Explainable Artificial Intelligence (MDMIoV-DLXAI). The main intention of the MDMIoV-DLXAI model is to enhance the malware detection and classification model in IoV by utilizing advanced two-tier optimization models. Initially, the data normalization stage is performed by the min-max normalization to convert input data into a beneficial format. Besides, the proposed MDMIoV-DLXAI model utilizes the reptile search algorithm (RSA) model for feature selection. Furthermore, the hybrid of bidirectional long short-term memory with a multi-head self-attention (BiLSTM-MHSA) model is employed for the malware classification process. The parameter tuning process is performed through the pelican optimization algorithm (POA) to improve the classification performance of the BiLSTM-MHSA classifier. Finally, SHAP is utilized as an XAI technique to enhance malware detection and decision-making processes of AI-driven security systems. The experimental evaluation of the MDMIoV-DLXAI method is examined under the malware dataset. The comparison study of the MDMIoV-DLXAI method demonstrated a superior accuracy value of 97,393% over existing techniques.

Background: The epidemiological pattern of hepatitis A infection has shown dynamic changes in many parts of the world due to improved socio-economic conditions and the accumulation of seronegative subjects, which leads to possible outbreaks and increased morbidity rate. In Tunisia, the epidemiological status of hepatits A virus is currently unknown. However, over the past years higher numbers of symptomatic hepatitis A virus infection in school attendants and several outbreaks were reported to the Ministry of Health, especially from regions with the lowest socio-economic levels in the country. The aim of this study was to investigate the current seroprevalence of hepatitis A virus antibodies in central-west Tunisia and assess the impact of hepatitis A virus vaccination on hepatitis A epidemiology. Methods: Serum samples from 1379 individuals, aged 5–75 years, were screened for hepatitis A virus antibodies. Adjusted seroprevalence, incidence and force of infection parameters were estimated by a linear age structured SEIR (Susceptible-Exposed-Infectious-Recovered) compartmental model. A vaccine model was then constructed to assess the impact on hepatitis A virus epidemiology of 3 scenarios of vaccination strategies: one dose at 12-months of age, one dose at 6-years and one dose at 12-months and another at 6-years of age during 6 years. Results: A rapid increase in anti-hepatitis A virus seroprevalence was noted during infancy and adolescence: 47% of subjects under 10-years-old are infected; the prevalence increases to 77% at 15-years and reaches 97% in subjects aged 30-years. The force of infection is highest between 10 and 30-years of age and the incidence declines with increasing age. The vaccine model showed that the 3-scenarios lead to a significant reduction of the fraction of susceptibles. The two doses scenario gives the best results. Single-dose vaccination at 6-years of age provides more rapid decrease of disease burden in school-aged children, as compared to single-dose vaccination at 12-months, but keeps with a non-negligible fraction of susceptibles among children < 6-years. Conclusions: Our study confirms the epidemiological switch from high to intermediate endemicity of hepatitis A virus in Tunisia and provides models that may help undertake best decisions in terms of vaccinations strategies.