Current Organizations / Institutes

The organizations/institutes from the scholar's latest article

Previous Organizations / Institutes

The organizations / institutes from the scholar's other articles in the past ten years


Pedro César Santana Mancilla has a total of 161 co-authors. Here are the top co-authors by number of shared publications:


Latest Publications

  • Open Access

    Beyond Grades: Temporal Validation of a Multi-Output Model for Formative Academic Indicators in a Latin American Public University

    by Miguel Ángel Rodríguez-Ortiz, Pedro C. Santana-Mancilla, Luis E. Anido-Rifón

    Learning-platform records are often assembled after a course ends, obscuring whether predictors were available at an earlier decision point. We reconstructed student-course-period observations from pseudonymized assignment, assessment, learning-management-system, and forum events across nine official undergraduate periods (2020-1 to 2024-1) at a Mexican public university. For 7963 candidate observations, 16 predictors used events at or before 25%, 50%, or 75% of each period; five formative outcomes were defined on assignments due and examinations started after the cutoff. A shared-trunk multi-output neural network was evaluated with grouped cross-validation, course-held-out validation, and a retrospective 2024-1 holdout. Only future assignment non-submission showed a consistent aggregate signal (grouped cross-validation R2 = 0.217, 0.253, and 0.147). At 50%, leave-one-course-out R2 was 0.211, but median course-level R2 was −0.146. On the 2024-1 holdout, R2 was 0.144 (95% confidence interval [−0.328, 0.335]; n = 471), below a prior-submission persistence rule (R2 = 0.324). Other outcomes lacked stable performance, and prior examination accuracy approximated future accuracy better than the model. The findings support a limited retrospective association, not reliable course-level prediction, early-warning deployment, or intervention effectiveness. The contribution is a temporally explicit reconstruction and validation workflow for an underrepresented Latin American context.

  • Open Access

    Tutoría universitaria y permanencia escolar: evidencia longitudinal del Programa Institucional de Tutorías de la Universidad de Colima (2015–2024)

    by Edgar Alfredo Nande Vázquez, Pedro César Santana Mancilla

    El estudio analiza el Programa Institucional de Tutorías (PIT) de la Universidad de Colima mediante un enfoque cuantitativo longitudinal, con el propósito de evaluar su relación con la permanencia escolar, la eficiencia terminal y la titulación. Para ello, se utilizaron datos administrativos de 30 facultades, observadas anualmente entre 2015 y 2024, mediante modelos estadísticos de datos de panel con efectos fijos. Los resultados muestran que la tutoría universitaria es una estrategia institucional relevante para fortalecer las trayectorias académicas; sin embargo, sus efectos no son automáticos ni homogéneos. Su impacto depende de la coherencia del modelo tutorial, la continuidad del acompañamiento y la focalización de las intervenciones. Asimismo, se identificó que los factores relacionados con la permanencia temprana difieren de aquellos vinculados con la eficiencia terminal, por lo que es necesario implementar estrategias diferenciadas según el momento de la trayectoria académica. El estudio aporta evidencia longitudinal para fortalecer la tutoría como estrategia institucional orientada a favorecer la permanencia, la progresión académica y la culminación de los estudios universitarios.

  • Open Access

    TinyStressNet: A Quantization-Ready Model for Subject-Independent Academic Stress Sensing

    by Pablo A. Alcaraz-Valencia, Pedro C. Santana-Mancilla, Laura S. Gaytán-Lugo, Luis Anido-Rifón

    TinyStressNet is a compact neural classifier for three-class academic stress estimation (low, medium, high) from three physiological features (galvanic skin response, heart rate, and skin temperature) designed for low-cost educational sensing. We evaluate it on 1000 interval-level observations from five engineering students collected over three months, with labels derived from repeated 6-item short-form of the State-Trait Anxiety Inventory (STAI-6) self-reports aligned with contemporaneous physiological readings. For subject-independent evidence, we use leave-one-subject-out (LOSO) evaluation as the primary protocol, complemented by a per-user chronological split for temporal robustness. We study augmentation through the lens of approximate invariance to bounded, label-preserving nuisance transformations, comparing no augmentation, Gaussian jitter, a radial-rescaling family, and a bounded affine (BA) family against strong classical baselines (Linear Discriminant Analysis (LDA), Quadratic Discriminant Analysis (QDA), and Support Vector Machines (SVMs), logistic and ordered-logit models). Under LOSO, the best TinyStressNet configuration reaches 0.899 accuracy and 0.898 weighted F1; within users, it reaches 0.912 accuracy. The model uses 3139 parameters and 2800 Multiply-Accumulate operations (MACs), and post-training int8 export preserves predictions with almost the same accuracy (0.992 float–int8 agreement). Strong classical baselines remain competitive on this low-dimensional representation, which we read as evidence about the geometry of the three-feature space rather than as a limitation of the model. Repeating the full evaluation under ten seeds shows that the augmentation families are not separable on this representation: subject-independent weighted F1 spans 0.894 to 0.900 across the four conditions with a seed-to-seed standard deviation of about 0.005, and no augmentation is not measurably worse than the best family. The BA family is the most consistent condition, the strongest in five of ten seeds and never the weakest, which we report as a stability observation rather than a performance gain. TinyStressNet is presented as a compact, quantization-ready reference model and a controlled testbed for invariance-aware augmentation in low-dimensional physiological sensing.

  • Open Access

    Beyond Grades: Multi-Target Deep Learning for Early Academic Risk Detection

    by Miguel Ángel Rodríguez-Ortiz, Pedro C. Santana-Mancilla, Luis E. Anido-Rifón

    Early identification of academic risk remains a major challenge in public higher education, where institutional data are often fragmented and underused. This study proposes a multi-output deep learning model to predict five formative indicators during the semester: assignment performance, missing assignments, procrastination rate, exam performance, and exam accuracy. The dataset included 4,230 student-course records and 26 features extracted from learning and assessment platforms. A multitask regression architecture with shared and task-specific layers was optimized through Bayesian search and regularization. The model achieved strong predictive performance (R² = 0.9803; MAE = 0.0239; RMSE = 0.0404), outperforming traditional machine learning baselines. SHAP analysis identified temporality and engagement variables as the most influential predictors. Additional ordinal and clustering analyses revealed engagement profiles useful for early feedback and support. Results highlight the value of multi-output learning analytics for timely and privacy-aware academic intervention.

  • Open Access

    Hacia robots que conversan: mejorando la Interacción Humano-Robot con Inteligencia Artificial generativa

    by David Alonso Flores Castro, Joel Adrián García Bravo, Jorge Gudiño-Lau, Pedro C. Santana-Mancilla

    El artículo presenta el diseño y evaluación de un sistema de Interacción Humano-Robot (HRI) basado en lenguaje natural, cuyo objetivo es acercar la robótica educativa a una experiencia más conversacional e intuitiva. Se desarrolló un prototipo utilizando el robot Yahboom DOFBOT, integrando reconocimiento de voz local (Vosk), interpretación semántica mediante un modelo de lenguaje multimodal (Gemini) y control cinemático ejecutado en una Jetson Nano bajo un enfoque de Interfaz de Lenguaje Natural para Robótica (NLIR). El sistema fue evaluado con 21 estudiantes universitarios, quienes debían solicitar siete acciones básicas usando expresiones libres. Se obtuvo una tasa general de acierto del 74.8%, destacando mayor precisión en comandos directos y dificultades en instrucciones ambiguas o graduales, especialmente en posiciones extremas. El 25.2% de errores se distribuyó principalmente en falta de contexto (43.2%) e interpretación del LLM (32.4%). A pesar de estas limitaciones, la percepción de los usuarios fue altamente positiva, con un índice global de interacción de 4.35/5. Los resultados muestran que la integración de LLMs en robots físicos permite interacciones más naturales y accesibles, aunque persisten desafíos en ambigüedad semántica y latencia. El estudio evidencia el potencial de los robots conversacionales en entornos educativos.

  • Gamifying Cybersecurity Education

    by María del Rocío Ramirez Jimenez, Elba A. Morales-Vanegas, Miguel A. Garcia-Ruiz, Pedro Damian Reyes, Pedro C. Santana-Mancilla

    The rapid digital transformation of higher education demands innovative strategies to address the growing challenges of cybersecurity. Universities must prepare students not only with technical skills but also with the ability to act responsibly in interconnected digital environments. Gamification has emerged as a powerful pedagogical approach to engage learners, foster motivation, and translate complex concepts into meaningful experiences. This chapter introduces a gamified learning platform designed to raise cybersecurity awareness and promote digital responsibility among university students. The platform integrates interactive scenarios, self-assessment tools, and educational games to develop essential skills such as critical thinking, decision-making, and ethical digital behavior. Early results suggest its potential to strengthen students' understanding of digital risks and encourage safer online practices. Future iterations envision adaptive content, AI-driven personalization, and advanced analytics to enhance impact.

  • Co-Designing a Serious Game for Early Literacy

    by Laura S. Gaytán-Lugo, Pedro C. Santana-Mancilla, Pablo A. Alcaraz-Valencia, Eduardo D. Jimenez Cisneros, Cristina Zéster Pelayo

    Video games are increasingly recognized as engaging tools to enhance learning. In Mexico, where children's reading performance presents challenges, serious games offer innovative opportunities to foster early literacy. This study aimed to design and evaluate a serious game to support reading and writing skills in preschool children. Using a participatory design approach, teachers and children actively co-created the proposal to ensure relevance and acceptance. The game was developed with the Unity platform and integrated three activities: completing words, matching letters with images, and a memory game. Evaluation sessions were carried out with children at a support school, complemented by teacher feedback. Results revealed strong acceptance of the technology, reflected in children's motivation, enjoyment, and sustained attention. Suggestions included increasing difficulty levels and incorporating phonetic feedback, providing insights for improving future iterations, and expanding educational impact.

  • Open Access

    Sustainable Urban Mobility: Leveraging Generative AI for Symmetry-Aware Traffic Light Optimization

    by Pedro C. Santana-Mancilla, Antonio Guerrero-Ibáñez, Juan Contreras-Castillo, Jesús García-Mancilla, Luis Anido-Rifón

    Urban intersections are critical nodes where traffic congestion and energy inefficiency converge. Traditional signal control systems often optimize either mobility or sustainability, creating an asymmetry between flow efficiency and environmental impact. This study introduces a symmetry-aware generative optimization framework that leverages Generative Artificial Intelligence (GAI) to balance both dimensions. Using the microscopic simulator SUMO, we modeled a signalized intersection in Colima, Mexico, under five control strategies: Fixed Time (baseline), GPT-4o, GPT-5 Thinking, Gemini 2.5 Pro, and DeepSeek V3. Each Large Language Model (LLM) received structured simulation data and generated new phase-duration configurations to minimize queue length, travel time, and CO2 emissions while improving average speed. Step-level performance was evaluated using descriptive statistics, and Wilcoxon signed-rank tests paired with Holm–Bonferroni correction. Results show that all LLM-based controllers significantly outperformed the Fixed Time baseline (adjusted p ≤ 4.8 × 10−6), with large effect sizes (|dz| ≈ 1.5–2.6). GPT-5 achieved the strongest performance, reducing queue size by ≈ 44%, CO2 emissions by ≈ 17%, and increasing average speed by ≈ 58%. The results validate the feasibility of symmetry-aware generative reasoning for sustainable traffic optimization and establish a reproducible methodological framework applicable to future AI-driven urban mobility systems.


  • Open Access

    Extended Reality in Computer Science Education: A Narrative Review of Pedagogical Benefits, Challenges, and Future Directions

    by Miguel A. Garcia-Ruiz, Elba A. Morales-Vanegas, Laura S. Gaytán-Lugo, Pablo A. Alcaraz-Valencia, Pedro C. Santana-Mancilla

    Technologies such as XR (Extended Reality), in the form of VR (Virtual Reality), AR (Augmented Reality) and MR (Mixed-Reality), are being researched for their potential to support higher education. XR offers novel opportunities for improving understanding and engagement of computer science (CS) courses, abstract and algorithmic thinking and the application of knowledge to solve problems with computers. This narrative literature review aims to report the state of XR adoption in the university CS education context by studying pedagogical benefits, representative cases, challenges, and future research work. Recent case studies have demonstrated that VR innovations are supportive of algorithm and data structure visualization, AR in programming and circuit analysis contextualization, and MR in bridging the experimental practice on virtual with real hardware within computer labs. The potential of XR to enhance engagement, motivation, and complex content understanding has already been researched. However, ongoing obstacles remain such as the high cost of hardware, technical issues in practicing scalable content, restricted access for students with disabilities, and ethical considerations over privacy and data protection. This review also presents XR, not as a substitute for traditional pedagogy, but as an additive tool that, in alignment with well-defined curricular objectives, may enhance CS learning. If it overcomes these deficiencies and progresses appropriate inclusive evidence-based practices, XR has the potential to play a powerful role in the future of computer science education as part of the digital learning ecosystem.


  • Open Access

    Towards Human-Centered Assisted Mobility in Intelligent Road Environments

    by Juan Guerrero-Ibañez, Pedro C. Santana-Mancilla, Juan Contreras-Castillo

    In this paper, we explore how Human-Centered Artificial Intelligence (HCAI), the Internet of Everything (IoE), and human-centric 6G ecosystems can converge to improve assisted road mobility and accessibility for vulnerable road users. Our study introduces a conceptual framework that integrates intelligent vehicles, road infrastructure, and pedestrians into a unified, inclusive, and interactive environment. By emphasizing human-centered design principles, the proposal highlights how emerging technologies can enhance user experience, safety, and autonomy for people with disabilities and other road users. Through the analysis of representative use cases, we discuss how HCAI and real-time communication technologies can enable adaptive interaction, personalized navigation, and inclusive participation in future mobility systems. The results outline a research agenda for developing intelligent road environments where technology actively supports human needs and values.