Abstract
Lung cancer is one of the leading causes of cancer-related deaths worldwide, mainly due to late diagnosis. Screening programs can benefit from Computer-Aided Diagnosis (CAD) systems that detect and classify lung nodules using Computed Tomography (CT) scans. A great proportion of the literature proposes deep learning models based on single and private datasets with no evaluation of their generalisation capability. The main goal of this work is to study and address the lack of generalisation to out-of-domain data (source domain different from the target domain). In this work, we propose using a ResNet architecture with 2.5D inputs capable of maintaining the spatial information of the nodules (3 input channels based on the anatomical planes). Secondly, we apply domain-specific data augmentation tailored for CT scans. Combined with data augmentation, using 2.5D inputs achieves the best results, both in in-domain data (LIDC-IDRI: N=1377 nodules; and LNDb: N=183 nodules) and in out-of-domain data (LUNGx: N=73 nodules). In in-domain data, an Area Under the Curve (AUC) of 0.914 was achieved in the internal test set and 0.746 in one of the external test sets. Notably, in out-of-domain data, where the ground-truth labels have been confirmed by biopsy, whereas the training data only involved radiologist annotation regarding the likelihood of malignancy, AUC improves from 0.576 to 0.695, reaching a performance close to that of radiology experts. In the future, strategies should be applied to deal with the level of uncertainty of lung nodule annotations based solely on the observation of the CT scans.

Abstract
The integration of solar photovoltaic (PV) systems and smart grids has enabled distributed energy trading, yet the development of regulatory frameworks for microgrid energy markets remains a challenge. Rising energy costs and greenhouse gas emissions necessitate innovative strategies to ensure affordable, sustainable, and reliable power for communities. This paper proposes a Community Energy Market (CEM) leveraging Linear Programming (LP) optimization to minimize energy costs and enhance renewable energy utilization. The results demonstrate that the CEM approach significantly increases energy self-sufficiency, reducing reliance on the grid. This method achieves Rs.38,830 cost saving. Furthermore, local energy trading within communities yields 68.75% % energy savings and reduces CO2 emissions by 88.01%. These findings highlight the effectiveness of the CEM model in fostering community collaboration, improving microgrid resilience, and promoting environmental sustainability. The proposed solution emphasizes the need for diversifying energy sources and adopting advanced energy market systems to deliver long-term, cost-effective, and eco-friendly energy solutions.

Abstract
This workshop proposal is the third edition of a workshop which has been organised at EICS 2023 and EICS 2024. This edition aims to bring together researchers and practitioners interested in the engineering of interactive systems that embed AI technologies (as for instance, AI-based recommender systems) or that use AI during the engineering lifecycle. The overall objective is to identify (from experience reported by participants) methods, techniques, and tools to support the use and inclusion of AI technologies throughout the engineering lifecycle for interactive systems. A specific focus is on guaranteeing that user-relevant properties such as usability and user experience are accounted for. Contributions are also expected to address system-related properties, including resilience, dependability, reliability, or performance. Another focus is on the identification and definition of software architectures supporting this integration.

Abstract
The development of artificial intelligence (AI) in the last decade has reshaped government operations and raised privacy concerns as automated processes become commonplace. This study aims to identify the main privacy issues associated with government use of AI in public services. Using a bibliometric analysis that includes co-citation of references and authors, bibliographic coupling, and keyword co-occurrence approaches, the study analyzed the literature on this topic through VOSViewer and the Web of Science database. Findings highlight significant privacy concerns: (i) opaque data-driven decisions, (ii) bias in predictive algorithms, (iii) difficulty obtaining explanations for decisions, (iv) mistrust in AI systems, (v) ethical lapses in AI execution, and (vi) trust deficit in government AI use. Additionally, 18 research questions are defined, addressing ethical limits of privacy in AI government use. A consensus in the literature urges governments to enact laws ensuring data privacy "by default" in AI decision-making and data management/transfer to third parties. © The Editor and Contributing Authors Severally 2025. All rights reserved.

Abstract
This paper introduces a passive acoustic method for detecting and tracking marine vessels via Time Difference of Arrival (TDoA) estimates collected by an array of synchronized Intelligent Buoys (SIBs). A 24-hour deployment recorded a passing tanker's acoustic signature, which we processed with band-pass filtering and sliding-window cross-correlation, to extract robust TDoA time series. We implemented a nonlinear Gauss-Newton estimator to reconstruct the vessel's trajectory. Position tracking fails, given the geometric configuration of the SIBs with regard to the vessel's trajectory but we suggest a possible solution to overcome this problem using synthesized data inspired on the experiment.

Abstract
Ensuring digital resources are accessible to all users, including those with disabilities, is critical in today's digital landscape. The growing volume of online content has intensified the need for automated accessibility evaluations to ensure compliance with accessibility guidelines. Yet, existing automated tools are limited in scope, being unable to identify many types of accessibility barriers. Recent advances in AI, particularly large language models (LLMs), offer opportunities to expand the range of automated accessibility checks. This work explores the ability of LLMs to detect accessibility barriers related to web page headings. We developed targeted prompts to help LLMs identify them and evaluated the effectiveness of three models - Llama 3.1, GPT-4o, and GPT-4o mini - in multiple versions of a reference webpage, each featuring different heading-related barriers. Findings reveal that model performance depends on barrier type: Llama 3.1 stands out at detecting structural issues like heading appropriateness and hierarchy, GPT-4o is better at identifying accessible names and semantic substitutions, while GPT-4o mini is the most versatile, handling complex structural modifications and labelling. This study highlights LLM's potential in advancing web accessibility evaluation and bridging gaps in automated assessments.