Abstract

Abstract
Algorithmic bias in healthcare mirrors existing data biases. However, the factors driving unfairness are not always known. Medical devices capture significant amounts of data but are prone to errors; for instance, pulse oximeters overestimate the arterial oxygen saturation of darker-skinned individuals, leading to worse outcomes. The impact of this bias in machine learning (ML) models remains unclear. This study addresses the technical challenges of quantifying the impact of medical device bias in downstream ML. Our experiments compare a “perfect world”, without pulse oximetry bias, using SaO2 (blood-gas), to the “actual world”, with biased measurements, using SpO2 (pulse oximetry). Under this counterfactual design, two models are trained with identical data, features, and settings, except for the method of measuring oxygen saturation: models using SaO2 are a “control” and models using SpO2 a “treatment”. The blood-gas oximetry linked dataset was a suitable test-bed, containing 163,396 nearly-simultaneous SpO2 - SaO2 paired measurements, aligned with a wide array of clinical features and outcomes. We studied three classification tasks: in-hospital mortality, respiratory SOFA score in the next 24 h, and SOFA score increase by two points. Models using SaO2 instead of SpO2 generally showed better performance. Patients with overestimation of O2 by pulse oximetry of = 3% had significant decreases in mortality prediction recall, from 0.63 to 0.59, P < 0.001. This mirrors clinical processes where biased pulse oximetry readings provide clinicians with false reassurance of patients’ oxygen levels. A similar degradation happened in ML models, with pulse oximetry biases leading to more false negatives in predicting adverse outcomes. © The Author(s), under exclusive license to Springer Nature Switzerland AG 2025.

Abstract
Artificial I ntelligence ( AI) a nd M achine Learning are frequently used to develop player skills in robotic soccer scenarios. Despite the potential of deep reinforcement learning, its computational demands pose challenges when learning complex behaviors. This work explores less demanding methods, namely Evolution Strategies (ES) and Hierarchical Reinforcement Learning (HRL), for enhancing coordination and cooperation between two agents from the FC Portugal 3D Simulation Soccer Team, in RoboCup. The goal is for two robots to learn a high-level skill that enables a robot to pass the ball to its teammate as quickly as possible. Results show that the trained models under-performed in a traditional robotic soccer two-agent task and scored perfectly in a much simpler one. Therefore, this work highlights that while these alternative methods can learn trivial cooperative behavior, more complex tasks are difficult t o learn.

Abstract
Incidental visualizations are meant to be perceived at-a-glance, on-the-go, and during short exposure times, but are not seen on demand. Instead, they appear in people's fields of view during an ongoing primary task. They differ from glanceable visualizations because the information is not received on demand, and they differ from ambient visualizations because the information is not continuously embedded in the environment. However, current graphical perception guidelines do not consider situations where information is presented at specific moments during brief exposure times without being the user's primary focus. Therefore, we conducted a crowdsourced user study with 99 participants to understand how accurate people's incidental graphical perception is. Each participant was tested on one of the three conditions: position of dots, length of lines, and angle of lines. We varied the number of elements for each combination and the display time. During the study, participants were asked to perform reproduction tasks, where they had to recreate a previously shown stimulus in each. Our results indicate that incidental graphical perception can be accurate when using position, length, and angles. Furthermore, we argue that incidental visualizations should be designed for low exposure times (between 300 and 1000 ms).

Abstract
Palaeoclimate time series, reflecting the state of Earth's climate in the distant past, occasionally display very large and rapid shifts showing abrupt climate variability. The identification and characterisation of these abrupt transitions in palaeoclimate records is of particular interest as this allows for understanding of millennial climate variability and the identification of potential tipping points in the context of current climate change. Methods that are able to characterise these events in an objective and automatic way, in a single time series, or across two proxy records are therefore of particular interest. In our study the matrix profile approach is used to describe Dansgaard-Oeschger (DO) events, abrupt warmings detected in the Greenland ice core, and Northern Hemisphere marine and continental records. The results indicate that canonical events DO-19 and DO-20, occurring at around 72 and 76 ka, are the most similar events over the past 110 000 years. These transitions are characterised by matching transitions corresponding to events DO-1, DO-8, and DO-12. They are abrupt, resulting in a rapid shift to warmer conditions, followed by a gradual return to cold conditions. The joint analysis of the delta 18O and Ca2+ time series indicates that the transition corresponding to the DO-19 event is the most similar event across the two time series.

Abstract
The notable expansion of technologies related to automated processes has been observed in recent years, largely driven by the significant advantages they provide across diverse industries. Concurrently, there has been a rise in simulation technologies aimed at replicating these complex systems. Nevertheless, in order to fully leverage the potential of these technologies, it is crucial to ensure the highest possible resemblance of simulations to real-world scenarios. In brief, this work consists of the development of a data acquisition and processing pipeline allowing a posterior search for the optimal physical parameters in MuJoCo simulator to obtain a more accurate simulation of a dexterous robotic hand. In the end, a Random Search optimization algorithm was used to validate this same pipeline.