Abstract

Abstract

Abstract
Reducing office buildings’ energy consumption can contribute significantly towards carbon reduction commitments since it represents 10% of total energy consumption. Major components are lighting (40% of consumption), electrical equipment (35%), and heating and central cooling systems (25\%). Occupants’ behaviours impact these energy consumption components, with solid evidence on the role of individual behaviours. In this work, we propose a methodology that uses virtual choreographies to identify and prioritize behaviour-change interventions towards office users based on the potential impact on energy consumption. The data shows that some behaviours with significant consumption have little potential for behavioural change impact, while other behaviours hold substantial potential for lowering energy consumption via behavioural change.

Abstract
Peripheral arterial disease (PAD) causes blockage of the arteries, altering the blood flow to the lower limbs. This blockage can cause the individual with PAD to feel severe pain in the lower limbs. The main contribution of this research is the discovery of a solution that allows the automatic detection of the onset of claudication based on data analysis from patients' smartphones. For the data-collection procedure, 40 patients were asked to walk with a smartphone on a thirty-meter path, back and forth, for six minutes. Each patient conducted the test twice on two different days. Several machine learning models were compared to detect the onset of claudication on two different datasets. The results suggest that we can identify the onset of claudication using inertial sensors with a best case accuracy of 92.25% for the Extreme Gradient Boosting model.

Abstract
Large Language Models (LLMs) are widely used today in virtual assistants and content generation. However, there are suspicions that LLMs present confirmation bias, responding in a way that reinforces beliefs or assumptions embedded in users' questions, which can lead to erroneous decision-making, especially in sensitive areas such as healthcare. The objective of this research is to determine how often and under what conditions LLMs present confirmation bias and to identify the causes of this effect. The methodology involves conducting an experiment in which 52 biased healthcare questions are presented to 10 of the most popular models and analyzing whether their responses were biased. This work proves with statistical power the behavior of confirmation bias. We show that confirmation bias in LLMs occurs in all LLMs with a frequency of 20% to 60% of the occasions. The evidence suggests that the bias arises from the training database, the Transformer architecture itself, and the instructions in the fine-tuning phase by the companies behind the LLMs. This research explores pathways for the development of trustworthy LLMs.

Abstract
The ongoing energy transition and the rapid electrification of transport increase the importance of integrating renewable energy sources into smart mobility systems. Among these, solar energy plays a central role, but the variability of solar radiation poses significant challenges for planning electric vehicle (EV) charging and ensuring the reliable operation of transport networks. This work addresses these challenges by combining Big Data approaches and High-Performance Computing (HPC) to improve solar radiation forecasting and assess its implications for sustainable transport as a novelty from previous works. A Long Short-Term Memory (LSTM) neural network was the focus, and it was trained to predict key meteorological variables - global solar radiation, temperature, and wind speed - using both the original dataset of 13 years and expanded datasets of up to 130 years, generated to simulate Big Data scenarios. Forecasting performance remained stable across datasets, with R2 values above 0.85 for all variables. The best predictive results were obtained for the original dataset, achieving R2 = 0.9884 for solar radiation, while the HPC reduced execution time compared to conventional desktop environments. These results demonstrate that larger datasets improve model scalability and robustness, but significantly increase computational demands. The Deucalion supercomputer achieved the best performance, processing the largest dataset (130 years) in 44.24 minutes, while the same task on a Ryzen 7 required 51.00 minutes. The proposed approach highlights the potential of integrating Big Data and HPC to support EV charging optimisation, smart grid operation, and sustainable mobility strategies, contributing to faster, more reliable, and data-driven decision-making in the energy-transport ecosystem. © 2025 IEEE.