Conference Agenda

This study provides a systematic literature review (SLR) complemented by a bibliometric analysis to examine the main trends and challenges in the integration of Artificial Intelligence (AI) in classroom learning and student performance between 2020 and 2025. Using the PRISMA protocol, 60 peer-reviewed articles were selected from Scopus, Web of Science, and PubMed databases. The results identify five key research lines: academic performance prediction, intelligent tutoring systems, psychological impacts of AI, automated assessment, and innovative applications such as gamification and robotics. The analysis highlights a strong focus on deep learning models, adaptive learning, and learning analytics, although a gap remains in pedagogical integration. Furthermore, student perceptions and ethical concerns emerge as critical factors influencing adoption. This review contributes to the academic field by offering a comprehensive synthesis of current literature and proposing a balanced integration of technological and pedagogical approaches to promote ethical, effective, and sustainable AI use in education.

El estudio analiza el impacto de la Inteligencia Artificial de las Cosas (AIoT) en el sector retail a nivel global, utilizando una metodología mixta con encuestas a consumidores y entrevistas a líderes del sector. Se examina cómo el AIoT redefine las estrategias de ventas, el compromiso y la fidelización del cliente en distintos contextos internacionales. Los resultados revelan que el 69.4 % de los consumidores percibe que estas tecnologías mejoran la experiencia de compra y el 53.2 % estaría dispuesto a pagar más por una experiencia tecnológica eficiente, aunque un 28.4 % expresa preocupación por la privacidad de sus datos, con diferencias significativas por género y país. Entre los gerentes, el 60 % ha implementado AIoT, reportando mejoras en eficiencia operativa (80 %) y satisfacción del cliente (64.5 %), pese a barreras como altos costos y falta de personal capacitado. El análisis estadístico indica percepciones homogéneas por género y país en la mayoría de las variables, excepto en la privacidad. El estudio identifica patrones y disparidades entre mercados desarrollados y emergentes, ofreciendo recomendaciones para empresas, gobiernos y desarrolladores tecnológicos que buscan liderar la transformación digital del retail mediante innovación inteligente y responsable.

Abstract– The growing saturation of healthcare systems, due to operational overload and the complexity of data management, has driven interest in more efficient technological solutions. This systematic review aimed to identify how Deep Learning models have been applied in the medical field recently. Forty scientific articles indexed in Scopus, a database recognized for its high rigor and academic prestige, were analyzed, selected using the PRISMA protocol. The results showed that the most studied anatomical areas were the respiratory system (16%), endocrine (15%), and musculoskeletal (13%). The most frequently addressed clinical tasks included classification (47.5%) and prediction (32.5%), with recurring specialties such as oncology and radiology in both categories. The most frequently used model families were CNN (32.5%), ResNet (32.5%), and specialized models (32.5%), applied primarily to medical images. A growing interest was also identified in more advanced architectures and the use of diverse clinical data. These findings provide a current overview of the field and open the way to new opportunities to develop more scalable and adaptable solutions in the healthcare sector.

The Fourth Industrial Revolution has driven the adoption of Industry 4.0 (I4.0) in supply chains (SCs), enhancing agility, flexibility, transparency, and competitiveness while fostering Reconfigurable Supply Chains (RECs). However, the role of I4.0 in improving the supply chain (SC) and company performance (PERF) remains unclear. Using hierarchical regression analysis and a multiple mediation model with a sample of 309 international firms, we demonstrate that RECs are primarily cost-oriented, rather than structurally reconfigurable, with only six REC enablers significantly impacting performance. The results show that I4.0 fully mediates the relationship between REC and SC performance (SCP) and partially mediates the SCP-PERF link. Thus, I4.0 is evolving from a mediator to a potential moderator in the SCP-PERF relationship.

The automation of processes with artificial intelligence (AI) allows companies' operations to become more accurate and efficient, in this systematic literature review, an exhaustive analysis was made using the PRISMA method, the characteristics of automating a process were studied, It was found that automating a process in its abstract nature is inclined towards accuracy, robustness of the process as the model improves when processing new data and finally to efficiency, likewise an analysis of AI methods that are present in the literature in the last year (2024-205) was made, it was revealed that the most used methods are Machine learning, It was revealed that the most used methods are Machine learning, which receives and returns structured data, and deep learning being the most used and processing mostly unstructured input data and returning almost all structured data, it was also found that the family of deep learning models most present in the literature is YOLO and for machine learning the decision trees, it was found that the 4 most effective deep learning models, all with an effectiveness value of 100% are “Darknet-19”, “Resnet-18”, “Resnet-50” and “Resnet-10”, these were applied to the Health area and for Machine learning “Extra Trees” with 100% oriented to the health area being the area with more presence in the experiments with AI models to automate processes in the scientific literature.

Understanding driver activity in real time is complex, yet it is important for in-vehicle systems that aim to reduce car crashes. This work addressed the problem by relying on state-of-the-art methods, specifically, evaluating three convolutional neural network (CNN) models, such as: AlexNet, VGGNet and ResNet with the aim of predicting real-time driver activity and behaviour during driving. For the development, a dataset consisting of 10,751 images was used (The dataset was obtained from the Kaggle platform). To achieve good results, there are multiple factors, such as the volume of the dataset, the quality of the data, the application of optimisation techniques, among others. The findings of the proposal showed the VGGNet model to be the most efficient model for efficiently classifying and predicting driver’s behaviour, achieving an accuracy rate of over 98%. It is closely followed by the AlexNet model, with an accuracy rate of 98%, a very significant result for this type of task, this model also obtained an F1-Score of 86%, and a count rate of 75%. However, the metrics obtained by the ResNet model are much lower than the compassion of the other models, it only achieved an accuracy rate of 35.28%, which indicates that it has limitations in identifying features and predicting driver behaviour. Finally, it is concluded that the VGGNet model slightly outperforms the AlexNet model, which shows that both models are efficient for this type of task