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
The interaction between humans and technology has always been a key determinant factor of adoption and efficiency. This is true whether the interaction is with hardware, software or data. In the particular case of Information Retrieval (IR), recent developments in Deep Learning and Natural Language Processing (NLP) techniques opened the door to more natural and efficient IR means, no longer based on keywords or similarity metrics but on a distributed representation of meaning. In this paper we propose an agent-based architecture to serve as an interface with industrial systems, in which agents are powered by specific Large Language Models (LLMs). Its main goal is to make the interaction with such systems (e.g. data sources, production systems, machines) natural, allowing users to execute complex tasks with simple prompts. To this end, key aspects considered in the architecture are human-centricity and context-awareness. This paper provides a high-level description of this architecture, and then focuses on the development and evaluation of one of its key agents, responsible for information retrieval. For this purpose, we detail three application scenarios, and evaluate the ability of this agent to select the appropriate data sources to answer a specific prompt. Depending on the scenario and on the underlying model, results show an accuracy of up to 80%, showing that the proposed agent can be used to autonomously select from among several available data sources to answer a specific information need. © The Author(s), under exclusive license to Springer Nature Switzerland AG 2025.

Abstract
Many of today’s domains of application of Machine Learning (ML) are dynamic in the sense that data and their patterns change over time. This has a significant impact in the ML lifecycle and operations, requiring frequent model (re-)training, or other strategies to deal with outdated models and data. This need for dynamic and responsive solutions also has an impact on the use of computational resources and, consequently, on sustainability indicators. This paper proposes an approach in line with the concept of Frugal AI, whose main aim is to minimize the resources and time spent on training models by re-using models from a pool of past models, when appropriate. Specifically, we present and validate a methodology for similarity-based model selection in data streaming environments with concept drift. Rather than training a new model for each new block of data, this methodology considers a pool with only a subset of the models and, for each new block of data, will select the best model from the pool. The best model is determined based on the distance between its training data and the current block of data. Distance is calculated based on a set of meta-features that characterizes the data, and on the Bray-Curtis distance. We show that it is possible to reuse previous models using this methodology, leading to potentially significant saving of resources and time, while maintaining predictive quality. © The Author(s), under exclusive license to Springer Nature Switzerland AG 2025.

Abstract
Explainable AI (xAI) emerged as one of the ways of addressing the interpretability issues of the so-called black-box models. Most of the xAI artifacts proposed so far were designed, as expected, for human users. In this work, we posit that such artifacts can also be used by computer systems. Specifically, we propose a set of metrics derived from LIME explanations, that can eventually be used to ascertain the quality of each output of an underlying image classification model. We validate these metrics against quantitative human feedback, and identify 4 potentially interesting metrics for this purpose. This research is particularly useful in concept drift scenarios, in which models are deployed into production and there is no new labelled data to continuously evaluate them, becoming impossible to know the current performance of the model.

Abstract
In recent years, the EU has been pushing forward ground-breaking legislation that covers new digital environments and services, with a strong focus on Ethics and AI. This includes legislation such as the Artificial Intelligence Act, the Digital Services Act or the General Data Protection Regulation. This legislation is, however, often written in very general and high-level terms, leaving a lot of space for interpretation, and a gap concerning how it could or should be implemented, realistically. In this paper we look specifically at the principle of Transparency in the Digital Services Act. Specifically, we discuss the requirements concerning Transparency in the regulation, we identify the gaps, and propose concrete measures that can be considered to facilitate and guide its implementation.

Abstract
With the proliferation of ODR service providers, there is a critical necessity to establish mechanisms supporting their functioning, particularly while designing ODR processes. This article aims to examine the impact of process modelling using BPMN, and of its relevance in the integration of AI into ODR processes within the EU. BPMN allows a meticulous depiction of all the ODR process steps, stakeholders, and underlying data in structured formats that are readable and interpretable by both humans and AI, which enables its integration. The advantages include predictive analysis, identification of opportunities for continuous improvement, operational efficiency, cost and time reduction, and enhanced accessibility for self-represented litigants. Additionally, the transparency afforded by explicitly incorporating AI in BPMN notation fosters a clearer comprehension of processes, facilitating management and informed decision-making. Nevertheless, it remains imperative to address ethical concerns such as algorithmic bias, fairness, and privacy.