Abstract
Underwater exploration relies heavily on autonomous underwater vehicles and sensor platforms for sustained monitoring of marine environments, yet their operational duration is limited by energy constraints. To enhance energy efficiency, various control strategies have been proposed, including robust, optimal, and disturbance-aware approaches. Recent work introduced a variable structure controller (VSC) with a constant-amplitude control action for depth control of a platform equipped with a variable buoyancy module, achieving an average 22% reduction in energy use in comparison with conventional PID-based controllers. In a separate paper, the conditions for its closed-loop stability were proven. This study extends these works by proposing a controller with a variable-amplitude control action designed to minimize energy consumption. A formal proof of stability is provided to guarantee safe operation even under conservative assumptions. The controller is applied to a previously developed depth-regulated sensor platform using a validated physical model. Additionally, this study analyzes how the controller parameters and mission requirements affect stability regions, offering practical guidelines for parameter tuning. A method to estimate oscillation amplitude during hovering tasks is also introduced. Simulation trials validate the proposed approach, showing energy savings of up to 16% when compared to the controller using a constant-amplitude control action.

Abstract
In recent years, non-residential buildings have increasingly adopted renewable energy generation systems to align with the European Union's goal of achieving carbon neutrality by 2050. However, energy storage systems play a fundamental role in maximising the use of the generated renewable energy. Due to their high acquisition costs, adequately sizing these systems is essential. Moreover, applying an optimal scheduling strategy for energy storage operation can significantly improve the economic viability of such systems by reducing energy-related costs. In this paper, a MILP-based optimisation algorithm-incorporating battery lifespan constraints-is applied to a reference commercial building to schedule the operation of the storage system. A sensitivity analysis on the installed photovoltaic power and energy storage capacity is performed to evaluate their impact on the economic and operational performance of the optimisation algorithm under different sizing configurations.

Abstract
Arterial stiffness (AS) is a well-established predictor of cardiovascular events, including myocardial infarction and stroke. One of the most recognized methods for assessing AS is through arterial pulse wave velocity (aPWV), which provides valuable clinical insights into vascular health. However, its measurement typically requires specialized equipment, making it inaccessible in primary healthcare centers and low-resource settings. In this study, we developed and validated different machine learning models to estimate aPWV using common clinical markers routinely collected in standard medical examinations. Thus, we trained five regression models: Linear Regression, Polynomial Regression (PR), Gradient Boosting Regression, Support Vector Regression, and Neural Networks (NNs) on the EVasCu dataset, a cohort of apparently healthy individuals. A 10-fold cross-validation demonstrated that PR and NN achieved the highest predictive performance, effectively capturing nonlinear relationships in the data. External validation on two independent datasets, VascuNET (a healthy population) and ExIC-FEp (a cohort of cardiopathic patients), confirmed the robustness of PR and NN (R- (2)> 0.90) across different vascular conditions. These results indicate that by using easily accessible clinical variables and AI-driven insights, it is possible to develop a cost-effective tool for aPWV estimation, enabling early cardiovascular risk stratification in underserved and rural areas where specialized AS measurement devices are unavailable.

Abstract
Problem: Portable ECG/sEMG acquisition systems for telemedicine often lack application flexibility (e.g., limited configurability, signal validation) and efficient wireless data handling. Methodology: A modular biosignal acquisition system with up to 8 channels, 24-bit resolution and configurable sampling (1-4 kHz) is proposed, featuring per-channel gain/source adjustments, internal MUX-based reference drive, and visual electrode integrity monitoring; Bluetooth (R) transmits data via a bit-wise packet structure (83.92% smaller than JSON, 7.28 times faster decoding with linear complexity based on input size). Results: maximum 6.7 mu V-rms input-referred noise; harmonic signal correlations >99.99%, worst-case THD of -53.03 dBc, and pulse wave correlation >99.68% in frequency-domain with maximum NMSE% of 6e-6%; and 22.3-hour operation (3.3 Ah battery @ 150 mA). Conclusion: The system enables high-fidelity, power-efficient acquisition with validated signal integrity and adaptable multi-channel acquisition, addressing gaps in portable biosensing.

Abstract
Introduction: School lunches are essential in promoting a sustainable, healthy diet for children. However, students’ adherence in developed countries is modest. Objectives: A cross-sectional study with the objective of evaluating school lunch participation and exploring the factors influencing participation and its associations with sociodemographic characteristics, socioeconomic status, body weight, and adherence to the Mediterranean Dietary Pattern. Methodology: Baseline results from the R23 project, collected during the first five weeks of the 2022/23 school year, were used. In total, 806 students, aged 10–16 years, from the 2nd and 3rd cycles of basic education in Portugal, Benavente, participated in this study. Using a face-to-face structured questionnaire, students’ sociodemographic data, socioeconomic status, adherence to the Mediterranean Dietary Pattern, and school lunch participation were assessed. Daily school lunch participation was obtained via the school lunch booking service. Anthropometric measurements (weight, height and waist circumference) were performed on all students with signed consent forms. Results: The school lunch participation rate was 25.3%. Overall, 39.1% of students reported never attending the school canteen. Only 27% of the students eligible for Free or Reduced-Price Meals had an effective adherence to school lunch, and the non-participation rate in this group was 31%. Key barriers included living near to school and dissatisfaction with meal quality. Students' school lunch participation was significantly associated with age, grade, eligibility for Free or Reduced-Price Meals, and adherence to Mediterranean Dietary Pattern. Younger students and students eligible for Free or Reduced-Price Meals went to the canteen more often. While school lunch participation was positively and significantly associated with higher adherence to Mediterranean Dietary Pattern, no significant relationship was observed with body mass index or waist-to-height ratio. Students with greater adherence to Mediterranean Dietary Pattern displayed healthier body mass index and waist-to-height ratio. Conclusions: School lunch participation among adolescents was low, but increased participation was associated with higher adherence to the Mediterranean Dietary Pattern. These findings highlight the importance of school lunches in promoting healthier eating habits.

Abstract
Bringing AI models from digital to real-world applications presents significant challenges due to the complexity and variability of physical environments, often leading to unexpected model behaviors. We propose a framework that learns to translate images into control actions by modeling multimodal real-time data and system dynamics. This end-to-end controller offers enhanced explainability and robustness, making it well suited for complex manufacturing processes. This end-to-end framework differs from traditional approaches that rely on manually engineered features by learning complex relationships directly from raw data. Labels are only required during training to define the observable feature to be optimized. This adaptability significantly reduces development time and enhances scalability across varying conditions. This approach was tested in the Directed Energy Deposition (L-DED) process, a laser-based metal additive manufacturing technique that produces near-net-shape parts with exceptional energy efficiency and flexibility in both geometry and material selection. L-DED is inherently complex, involving multiphysics interactions, multiscale phenomena, and dynamic behaviors, which make modeling and optimization difficult. Effective control is crucial to ensure part quality in this dynamic environment. To address these challenges, we introduce Joint Embedding Multimodal Alignment with Sparse Identification of Nonlinear Dynamics for control (JEMA-SINDYc). It combines an image-based JEMA monitoring model, which predicts the melt pool size using only the on-axis sensor with labels provided by the off-axis camera, and dynamic modeling using SINDYc, which acts as a World Model by capturing system dynamics within the embedding space. Together, these components enable the development of an advanced controller trained via Behavioral Cloning. This approach improves part quality by minimizing porosity and reducing deformation. Thin-walled cylindrical parts were produced to validate and compare this approach with other control strategies, including both open-loop and JEMA-PID. This framework improves the reliability of AI-driven manufacturing and enhances control of complex industrial processes, potentially enabling wider adoption of the process.