Abstract
Background and Objectives: Early detection of cognitive impairment is crucial for timely clinical interventions aimed at delaying progression to dementia. However, existing screening tools are not ideal for wide population screening. This study explores the potential of combining machine learning, specifically, one-class classification, with simpler and quicker motor-cognitive tasks to improve the early detection of cognitive impairment. Methods: We gathered data on gait, fingertapping, cognitive, and dual tasks from older adults with mild cognitive impairment and healthy controls. Using one-class classification, we modeled the behavior of the majority group (healthy controls), identifying deviations from this behavior as abnormal. To account for confounding effects, we integrated confound regression into the classification pipeline. We evaluated the performance of individual tasks, as well as the combination of features (early fusion) and models (late fusion). Additionally, we compared the results with those from two-class classification and a standard cognitive screening test. Results: We analyzed data from 37 healthy controls and 16 individuals with mild cognitive impairment. Results revealed that one-class classification had higher predictive accuracy for mild cognitive impairment, whereas two-class classification performed better in identifying healthy controls. Gait features yielded the best results for one-class classification. Combining individual models led to better performance than combining features from the different tasks. Notably, the one-class majority voting approach exhibited a sensitivity of 87.5% and a specificity of 75.7%, suggesting it may serve as a potential alternative to the standard cognitive screening test. In contrast, the two-class majority voting failed to improve the low sensitivities achieved by the individual models due to the underrepresentation of the impaired group. Conclusion: Our preliminary results support the use of one-class classification with confound control to detect abnormal patterns of gait, fingertapping, cognitive, and dual tasks, to improve the early detection of cognitive impairment. Further research is necessary to substantiate the method's effectiveness in broader clinical settings.

Abstract
This article presents a real-time object detection and distance estimation system implemented on a low-cost platform. The system uses a Raspberry Pi 5 and two cameras in a stereoscopic configuration to capture pairs of images. Object detection is performed using YOLO neural networks and distance estimation is based on the disparity between the centers of the detected bounding boxes. The system is evaluated in terms of detection performance, inference speed and depth estimation accuracy. Three YOLO models (YOLOv8n, YOLO11n and YOLO12n) are tested at different resolutions. Among them, the YOLO11n with a resolution of 320×320 achieves the best balance between processing speed and detection quality in stereoscopic operation. The system has a low error in depth estimation at close range, with absolute errors of less than 1.2 cm up to 60 cm. At greater distances, accuracy is affected by the reduction in the size of the bounding box, which limits the reliability of the disparity. Possible improvements include using segmentation-based localization and optimizing the stereo configuration. The proposed system is suitable for short-range applications in controlled environments and serves as a basis for future improvements in embedded vision systems. © The Author(s), under exclusive license to Springer Nature Switzerland AG 2026.

Abstract
Parkinson's Disease (PD) is a neurodegenerative disease characterized by severe motor symptoms, with no cure to date, and deep brain stimulation (DBS) is one of the most effective therapies to reduce the symptoms. However, not all patients benefit equally of this therapy due to adverse effects caused by the stimulation of unwanted areas, making it essential to use diverse technologies when analysing the effect of the DBS in PD. Wearable devices, such as the iHandU System that analyses motor data combined with the Percept TM PC neurostimulator, capable of recording brain signals in real time, allow a more precise, personalized, and adaptive approach to treatment. While each method alone provides valuable but limited insights, combining and synchronizing data from the different sources enables a more comprehensive and dynamic understanding of the effects of stimulation on the patient. A study with 6 DBS-implanted participants from Centro Hospitalar Universitario de São João was conducted to test a synchronization protocol using both the Percept TM PC and the iHandU System. The protocol combined the Network Time Protocol and the Artifact-Based Synchronization Techniques, with an average of less than 500 ms of delay between signals. The results obtained show that this combination improves signal synchronization accuracy and consistency across subjects, minimizing delays and reducing reliance on visible peaks in cases of low signal quality or inconsistent artifact detection. © 2025 IEEE.

Abstract
The Internet of Vehicles (IoV) presents complex cybersecurity challenges, particularly against Denial-of-Service (DoS) and spoofing attacks targeting the Controller Area Network (CAN) bus. This study leverages the CICIoV2024 dataset, comprising six distinct classes of benign traffic and various types of attacks, to evaluate advanced machine learning techniques for instrusion detection systems (IDS). The models XGBoost, Random Forest, AdaBoost, Extra Trees, Logistic Regression, and Deep Neural Network were tested under realistic, imbalanced data conditions, ensuring that the evaluation reflects real-world scenarios where benign traffic dominates. Using hyperparameter optimization with Optuna, we achieved significant improvements in detection accuracy and robustness. Ensemble methods such as XGBoost and Random Forest consistently demonstrated superior performance, achieving perfect accuracy and macro-average F1-scores, even when detecting minority attack classes, in contrast to previous results for the CICIoV2024 dataset. The integration of optimized hyperparameter tuning and a broader methodological scope culminated in an IDS framework capable of addressing diverse attack scenarios with exceptional precision.

Abstract
The growing use of technology and social media has resulted in the emergence of digital influencers, a new profession capable of changing the mentalities and behaviours of those who follow them. This study arises to better understand the potential impact digital influencers might have on the Portuguese population's purchase behaviour and patterns, and for this purpose, seven hypotheses were formulated. An online questionnaire was conducted to respond to these theoretical assumptions and collected data from 175 respondents. A total of 129 valid answers were considered. It was possible to conclude that purchase intention does not necessarily translate into a purchase action. It was also concluded that the relationship between social network use and the purchase of products/services recommended by influencers is only statistically significant for Instagram. Furthermore, the individuals' generation is not statistically significant / linked with purchasing a product/service recommended by influencers. Yet further, a small percentage of respondents have also identified themselves as impulsive shoppers and perceived Instagram as their favourite social network. With the results of this study, it is also possible to state that the influencer's opinion was classified as the last factor considered in the purchase decision process. Additionally, there is a weak negative association between purchasing a product/service recommended by influencers with sponsorship disclosure and remunerated partnership, which decreases credibility and discourages purchasing, in Portugal, a feminine culture which dislikes materialism.

Abstract
Product design is a complex and iterative process that requires the balance of multiple constraints, such as material selection, manufacturability, regulatory compliance, and structural integrity, among others. Traditional design workflows follow a human-driven approach, limiting efficiency, adaptability, and the ability to quickly respond to evolving limitations. This paper introduces an agentic approach to product design, leveraging multi-agent systems to distribute and automate design tasks dynamically. To demonstrate this methodology, a hypothetical enclosure design is used as a guiding example, demonstrating how agents interact to generate product specifications, select materials, validate structural properties, assess manufacturability, and perform other relevant tasks throughout the design process. To implement this framework, CrewAI is utilized as an agent coordination system that enables the structured definition of roles and execution of tasks for autonomous agents. In the final section, a case study is presented, focusing on the design of a parallelepiped enclosure, applying the proposed framework in a simulated environment. Our findings highlight the advantages of agent-based collaboration in product design, showcasing its potential to optimize workflows, reduce development time, and improve adaptability to changing requirements. © The Author(s), under exclusive license to Springer Nature Switzerland AG 2026.