https://hdl.handle.net/10481/13709 2026-04-13T11:26:47Z https://hdl.handle.net/10481/112114 Assessing Bias in the Evaluation of Blood Glucose Prediction Models Rodríguez León, Ciro; Avilés-Pérez, María Dolores; Baños Legrán, Oresti; Lopez-Ibarra Lozano, Pablo J.; Muñoz Torres, Manuel Eduardo; Quesada Charneco, Miguel; Villalonga Palliser, Claudia Diabetes mellitus (DM) poses a critical global health challenge, with type 1 diabetes (T1D) patients presenting unique difficulties in maintaining a safe blood glucose level (BGL). This work demonstrates that evaluating BGL prediction models without considering different BGL ranges, hypoglycemia, hyperglycemia, and normoglycemia, introduces bias in assessing the prediction results. Data are obtained from the T1DiabetesGranada dataset, comprising over 22.5 million measured BGL values recorded at 15-min intervals, and are preprocessed into a uniform format for supervised learning. Time series are segmented into windows with a 2-h history length and prediction horizons of 30 and 60 min. An LSTM architecture is used to predict BGL values due to its ability to capture temporal dependencies. The evaluation combines traditional non-clinical metrics (RMSE, MAE, MAPE) with clinical metrics derived from the Clarke Error Grid. The newly proposed evaluation strategy assesses BGL prediction models performance not only across the entire BGL range but also within different BGL ranges. Results indicate that evaluation metrics computed using the entire BGL range may suggest satisfactory BGL prediction model performance. However, significant deficiencies emerge in hypoglycemic ranges, implying that conventional evaluation strategies may overestimate BGL prediction models capabilities. These findings highlight the need for a comprehensive evaluation strategy in different BGL ranges to avoid bias, especially while evaluating clinically critical regions. https://hdl.handle.net/10481/112037 Clustering type 1 diabetes patients based on blood glucose level measurements; Clustering type 1 diabetes patients based on blood glucose level measurements Rodríguez León, Ciro; Avilés-Pérez, M. D.; Baños, Oresti; Lopez-Ibarra Lozano, Pablo J.; Muñoz Torres, Manuel Eduardo; Quesada Charneco, Miguel; Villalonga Palliser, Claudia Improving diabetes mellitus management requires accurate blood glucose predictions to prevent dangerous glycemic extreme events; however, traditional models often struggle with the inherent fluctuations in blood glucose levels. In this work, the identification of distinct patient groups exhibiting common characteristics based on longitudinal continuous glucose monitoring (CGM) data is explored. This clustering approach aims to offer valuable insights for both enhancing predictive modeling accuracy and informing clinical relevance. Raw CGM data was transformed into structured feature sets using statistical descriptors calculated across clinically meaningful time intervals. Following outlier removal, multiple clustering algorithms were evaluated, with the optimal solution selected based on internal metrics and validation by clinical experts. The analysis revealed six distinct patient groups, each characterized by unique glycemic behaviors, including well-controlled, prone to hyperglycemia, and prone to hypoglycemia profiles. It is proposed that these identified clusters can improve glucose prediction by strategically balancing personalized and generalized approaches. Furthermore, valuable insights into individual metabolic variability can be obtained, potentially supporting the development of tailored treatment strategies. While acknowledging limitations inherent in data transformation and expert-driven evaluation, this clustering methodology represents a significant step towards more precise and data-driven diabetes mellitus management. https://hdl.handle.net/10481/112035 Exploring the Influence of Patients’ Heterogeneity on Automated Prediction of Blood Glucose Levels in Type 1 Diabetes Baños Legrán, Oresti; Avilés-Pérez, María Dolores; Gayaroa, Álvaro; Lopez-Ibarra Lozano, Pablo J.; Muñoz Torres, Manuel Eduardo; Quesada Charneco, Miguel; Rodríguez León, Ciro; Villalonga Palliser, Claudia This work investigates the impact of patients’ heterogeneity, including factors such as gender, age, and clinical conditions, on the performance of machine learning models in predicting blood glucose levels in individuals with Type 1 Diabetes. Using the T1DiabetesGranada dataset, various datasets were generated based on these heterogeneity factors. Three popular prediction models —a Linear model, an LSTM neural network, and a CNN— were applied to both the generated datasets and the complete dataset to measure prediction performance at a 30-minute prediction horizon. The preliminary results suggest that incorporating patient-specific heterogeneity factors generally improves prediction performance, highlighting the existence of bias in standard blood glucose level prediction models. Future research should explore whether these findings hold in other related datasets. This work has been supported by the PID2023-148188OA-I00 project ”RELIEF-T1D” which is funded by MICIU/AEI/10.13039/501100011033 and ERDF EU. https://hdl.handle.net/10481/111859 Event-based Vision for Early Prediction of Manipulation Actions Déniz Cerpa, José Daniel; Fermüller, Cornelia; Ros Vidal, Eduardo; Rodríguez Álvarez, Manuel; Barranco Expósito, Francisco Neuromorphic visual sensors are artificial retinas that output sequences of asynchronous events when brightness changes occur in the scene. These sensors offer many advantages including very high temporal resolution, no motion blur and smart data compression ideal for real-time processing. In this study, we introduce an event-based dataset on fine-grained manipulation actions and perform an experimental study on the use of transformers for action prediction with events. There is enormous interest in the fields of cognitive robotics and human-robot interaction on understanding and predicting human actions as early as possible. Early prediction allows anticipating complex stages for planning, enabling effective and real-time interaction. Our Transformer network uses events to predict manipulation actions as they occur, using online inference. The model succeeds at predicting actions early on, building up confidence over time and achieving state-of-the-art classification. Moreover, the attention-based transformer architecture allows us to study the role of the spatio-temporal patterns selected by the model. Our experiments show that the Transformer network captures action dynamic features outperforming video-based approaches and succeeding with scenarios where the differences between actions lie in very subtle cues. Finally, we release the new event dataset, which is the first in the literature for manipulation action recognition. This work was supported by the Spanish National Grant PID2019-109434RA-I00/ SRA (State Research Agency /10.13039/501100011033). We acknowledge the Telluride Neuromorphic Cognition Engineering Workshop (http: //www.ine-web.org), supported by NSF grant OISE 2020624 for the fruitful discussions on neuromorphic cognition and their participants for helping with the recording of the dataset. https://hdl.handle.net/10481/111854 Influence of Red and Blue Virtual Environments During a Listening Evaluation: A Preliminary Study of Academic Performance and Subjective Self-Perception Ávila Muñoz, Gabriel; López Gordo, Miguel Ángel; Rodríguez Álvarez, Manuel Color and lighting have been shown to influence human physiology and may affect cognitive performance. Some theories suggest that blue back-ground colors promote relaxation and concentration, whereas red colors induce arousal and alertness. However, empirical evidence remains inconclusive due to variations in experimental designs and task types, making it unclear how color ultimately affects cognitive performance. The aim of this preliminary study is to compare performance and subjective self-perception throughout a listening eval-uation and to determine whether red or blue colors yield differences in perfor-mance. We conducted a study in which university students completed five listen-ing tests within a red or blue virtual environment and answered questions about their subjective self-perception at the beginning and end of the experiment. Our results showed that: i) students in an environment characterized by a blue color obtained higher median performance (6.8 vs. 6.0) as time progressed during the evaluation; ii) medium positive correlations were found in the blue group (0.36 and 0.36) between the performance-attention and performance-relaxation that ex-ceeded that of the red group; iii) a medium negative correlation was discovered in the red group (-0.48) between the performance-mental effort that exceeded that of the blue group. The preliminary findings of this study could suggest that a blue environment may facilitate relaxation and attentional focus, which could enhance performance. Despite the modest sample size, these findings underscore the ne-cessity for additional research to determine the true impact of color on academic performance, with potential implications for educational settings and academic institutions. This work was supported by grant PID2021-128529OAI00, MCIN/AEI/https://doi.org/10.13039/501100011033, and by ERDF A way of making Europe; grant PROYEXCEL 00084, P21 00084, Projects for Excellence Research, funded by the Council for Economic Transformation, Industry, Knowledge and Universities, Junta de Andalucía 2021.