Abstract
The learning environment is an essential part of teaching and learning. Its personalization has several advantages (e.g., guaranteeing learning quality or effective learning). In vocational education, a personalized learning environment might provide training most suitable to each professional according to individual characteristics, skills, or career path. Artificial intelligence’s ability to process big data can be harnessed to personalize a learning environment. This work intends to investigate the personalization of a learning environment using artificial intelligence (AI) in vocational training that can provide relevant training based on the trainees’ skills required. A framework will be proposed to personalize a learning environment in this scope. Its development will follow the design science research (DSR) methodology. During the process, the survey methodology (expert interviews and focus groups) will be conducted to validate the artifact requirements and evaluate our future framework. © 2025 Elsevier B.V., All rights reserved.

Abstract
Cardiovascular diseases are the leading cause of death in the world, with coronary artery disease being the most prevalent. Coronary artery calcifications are critical biomarkers for cardiovascular disease, and their quantification via non-contrast computed tomography is a widely accepted and heavily employed technique for risk assessment. Manual segmentation of these calcifications is a time-consuming task, subject to variability. State-of-the-art methods often employ convolutional neural networks for an automated approach. However, there is a lack of studies that perform these segmentations with 3D architectures that can gather important and necessary anatomical context to distinguish the different coronary arteries. This paper proposes a novel and automated approach that uses a lightweight three-dimensional convolutional neural network to perform efficient and accurate segmentations and calcium scoring. Results show that this method achieves Dice score coefficients of 0.93 ± 0.02, 0.93 ± 0.03, 0.84 ± 0.02, 0.63 ± 0.06 and 0.89 ± 0.03 for the foreground, left anterior descending artery (LAD), left circumflex artery (LCX), left main artery (LM) and right coronary artery (RCA) calcifications, respectively, outperforming other state-of-the-art architectures. An external cohort validation also showed the generalization of this method's performance and how it can be applied in different clinical scenarios. In conclusion, the proposed lightweight 3D convolutional neural network demonstrates high efficiency and accuracy, outperforming state-of-the-art methods and showcasing robust generalization potential.

Abstract
Hypertension is the primary risk factor for cerebral small vessel disease (CSVD). However, its mechanistic links are yet to be completely understood. Advancements in diffusion-weighted magnetic resonance imaging (dMRI) increased sensitivity in detecting subtle white matter (WM) structural integrity changes. 44 hypertension patients without symptomatic CSVD underwent multi-modal evaluation of cerebral structure and function, including dMRI, neuropsychological tests and transcranial Doppler monitoring of the right middle cerebral artery (MCA) and left posterior cerebral artery (PCA) to assess neurovascular coupling (NVC). In the PCA, the modeled NVC curve was studied. We examined the cross-sectional relationship of WM integrity with NVC and cognitive performance, using correlational tractography. Diffusion measures from two dMRI models were used: fractional anisotropy, mean diffusivity, axial diffusivity, and radial diffusivity from diffusion tensor imaging, and quantitative anisotropy (QA) and isotropy from q-space diffeomorphic reconstruction. Regarding the NVC in the PCA, vascular elastic properties and initial response speed markers indicated better functional hyperemia with better WM integrity. However, the amplitude suggested increased NVC with worse WM integrity. In the MCA, increased NVC was associated with lower WM integrity. Better cognitive performance associated with preserved WM integrity. Increased functional hyperemia despite worse WM integrity may reflect less efficient NVC in hypertensive patients, potentially arising from (mal)adaptive mechanisms and brain network reorganization in response to CSVD. This observational study highlights the potential of transcranial Doppler and QA as susceptibility markers of pre-symptomatic CSVD.

Abstract
This work introduces Gen-JEMA, a generative approach based on joint embedding with multimodal alignment (JEMA), to enhance feature extraction in the embedding space and improve the explainability of its predictions. Gen-JEMA addresses these challenges by leveraging multimodal data, including multi-view images and metadata such as process parameters, to learn transferable semantic representations. Gen-JEMA enables more explainable and enriched predictions by learning a decoder from the embedding. This novel co-learning framework, tailored for directed energy deposition (DED), integrates multiple data sources to learn a unified data representation and predict melt pool images from the primary sensor. The proposed approach enables real-time process monitoring using only the primary modality, simplifying hardware requirements and reducing computational overhead. The effectiveness of Gen-JEMA for DED process monitoring was evaluated, focusing on its generalization to downstream tasks such as melt pool geometry prediction and the generation of external melt pool representations using off-axis sensor data. To generate these external representations, autoencoder (AE) and variational autoencoder (VAE) architectures were optimized using Bayesian optimization. The AE outperformed other approaches achieving a 38% improvement in melt pool geometry prediction compared to the baseline and 88% in data generation compared with the VAE. The proposed framework establishes the foundation for integrating multisensor data with metadata through a generative approach, enabling various downstream tasks within the DED domain and achieving a small embedding, allowing efficient process control based on model predictions and embeddings.

Abstract
This study investigated the relative contributions of medial temporal lobe (MTL) integrity and frontal lobe functions in behavioral variant frontotemporal dementia (bvFTD), Alzheimer's disease (AD), and mild cognitive impairment (MCI). We assessed 300 participants using neuropsychological memory tests-Free and Cued Selective Reminding Test (FCSRT), Brief Visuospatial Memory Test-Revised-and MTL volumetry. Propensity score matching examined which subtests were most influenced by frontal functions. Principal component analysis and ridge regression explored the relationship between MTL volumes and memory tests across groups. Significant differences in memory tests and MTL volumes followed the pattern: controls > MCI > bvFTD > AD. AD showed greater impairment than bvFTD in FCSRT Total (p = .013), Retention (p = .016), and Free Recall (p = .009). Correlations between MTL volumes and memory tests were strongest in MCI (r = -.58), followed by AD (r = -.36), controls (r = -.35), and non-significant in bvFTD (r = -.21). After propensity score matching, several group differences were no longer significant, including cued tasks, delayed recall, and retention. These findings support a continuum of frontal and MTL contributions to memory deficits. AD is primarily marked by MTL-related impairments, which are less pronounced in bvFTD. MTL volumetry influence on these memory tests in MCI underscores their utility in detecting subtle MTL-dependent memory dysfunction.

Abstract
This study examines the effectiveness of employing Advanced Rider Assistance System (ARAS) for enhancing motorcycle safety by reducing crashes and improving rider safety. The system includes both single solution approaches, like braking systems, and multi-sensor solutions that integrate data from LiDARs, radars, and cameras through sensor fusion. A systematic literature review was conducted to collect data from 2008 to 2024 across various sources related to ARAS. The review followed the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) guidelines to ensure a comprehensive and transparent process. Data were extracted from the included studies, focusing on study design, sample size, intervention details, and outcomes. The risk of bias was assessed using a customized checklist. The review included 31 studies that met the inclusion criteria. Findings were summarized for single sensor solutions and sensor fusion approaches.The review indicates that single-solution systems are effective ARAS technologies. In contrast, the application of sensor fusion in motorcycles has been only minimally explored, making it difficult to draw definitive conclusions about its impact in this context. Evidence from four-wheeled vehicles, however, shows that sensor fusion can enhance perception robustness, improve performance under adverse conditions, and contribute to measurable safety gains. These results suggest that similar advantages could be realized for motorcycles as fusion-based ARAS technologies become more widely implemented. Moreover, sensor fusion holds the potential to provide riders with broader situational awareness and more comprehensive safety assistance than single-system solutions. Future research should focus on addressing the identified challenges and optimizing these systems for broader implementation. This review underscores the critical role of ARAS in reducing motorcycle-related incidents and improving rider safety, highlighting the need for ongoing research to refine sensor fusion algorithms and address technical challenges for real-world applications.