Abstract
To maintain high levels of efficiency and compliance with delivery dates, automotive repair shops must have a good system for scheduling their activities. The scheduling of the activities of an automotive repair shop is a very complex task to be performed manually. Throughout this work, a Decision Support System (DSS) was developed and tested that considers two major constraints in an automotive workshop: human resources (technicians) and physical resources (work stalls). The proposed DSS has an embedded MIP model that assigns a technician and a work stall to each job, according to the input conditions. The DSS also generates schedules with the planning of technicians and jobs. The system was tested with real data from an automotive workshop and was able to create plans and schedules not only for the human and physical resources in but also to analyse the limiting resources of the workshop.

Abstract
Startups are vital to the economy and the digital future and are creators of around 50% of new jobs. Some studies indicate that around 90% of startups fail in their first months, mainly because they focus on launching products or services without adequate market validation. In addition, they have little or no experience in organisational management and lack the resources to apply quality models, which hinders their ability to face the challenges of a highly volatile and competitive environment. Therefore, this paper proposes the LightStartup framework, focused on startups in the service sector. LightStartup provides a lightweight, consistent and formalised process model, a process assessment model and a maturity model based on the ISO/IEC 33000 standard. LightStartup accompanies companies in transitioning from an informal management style to a formal and long-lasting management system, covering the management of services, people, customers and organisational governance.

Abstract

Abstract
Project NuClim (Nuclear observations to improve Climate research and GHG emission estimates) aims to use high-quality measurements of atmospheric radon activity concentration and ambient radioactivity to advance climate science and improve radiation protection and nuclear surveillance capabilities. It is supported by new metrological capabilities developed in the EMPIR project 19ENV01 traceRadon. This work reviews the scientific objectives of project NuClim in terms of both climate science and radiological protection, and provides an overview of the NuClim field campaign and the various nuclear measurements being implemented within the scope of the project. © S. Barbosa et al., Published by EDP Sciences, 2025.

Abstract
Pneumonia is a respiratory condition caused by various microorganisms, including bacteria, viruses, fungi, and parasites. It manifests with symptoms such as coughing, chest pain, fever, breathing difficulties, and fatigue. Early and accurate detection is crucial for effective treatment, yet traditional diagnostic methods often fall short in reliability and speed. Chest X-rays have become widely used for detecting pneumonia; however, current approaches still struggle with achieving high accuracy and interpretability, leaving room for improvement. PneumoNet, an artificial intelligence assistant for X-ray pneumonia detection, is proposed in this work. The framework comprises (a) a new deep learning-based classification model for the detection of pneumonia, which expands on the AlexNet backbone for feature extraction in X-ray images and a new head in its final layers that is tailored for (X-ray) pneumonia classification. (b) GPT-Neo, a large language model, which is used to integrate the results and produce medical reports. The classification model is trained and evaluated on three publicly available datasets to ensure robustness and generalisability. Using multiple datasets mitigates biases from single-source data, addresses variations in patient demographics, and allows for meaningful performance comparisons with prior research. PneumoNet classifier achieves accuracy rates between 96.70% and 98.70% in those datasets.

Abstract
The rapid proliferation of uncontrolled resources poses significant voltage regulation challenges in low-voltage (LV) distribution grids. In this condition, conventional charging strategies, often based on fixed or static schedules, may lead to adverse voltage deviations under unpredictable load conditions and variable renewable generation. To address these challenges, this paper studies a hybrid deep reinforcement learning (DRL) framework based on a Proximal Policy Optimization (PPO) policy network enriched by a Graph Convolution Variation (GCV) feature extractor to improve voltage regulation issues in LV grids. In addition to ensuring that electric vehicles (EVs) achieve their required state-of-charge (SoC), the framework dynamically adjusts charging rates in real time to maintain LV-grid voltage within acceptable limits. Extensive simulation results, including detailed analysis and comparisons with the static charging method, demonstrate significant improvements in voltage regulation, and enhanced overall grid performance. The obtained results demonstrate the effectiveness of controlling EVs' charging controls in an intelligent manner to address the voltage regulation issue in low-voltage grids. © 2025 Elsevier B.V., All rights reserved.