JVJuan Luna Valdez

Juan Luna Valdez

*This is an algorithmically generated author record based on publicly available information.

Current Organizations / Institutes

The organizations/institutes from the scholar's latest article


Juan Luna Valdez has a total of 25 co-authors. Here are the top co-authors by number of shared publications:


Latest Publications

  • Open Access

    Deep Learning Architecture Comparison for Pneumonia Detection in Pediatric Chest X-Rays: Ensemble Learning and Cross-Population External Evaluation

    by Jeanfranco Javier-Tarapaqui, Hugo Vega-Huerta, Luis Guerra-Grados, Jorge Pantoja-Collantes, Rubén Gil-Calvo, Oscar Benito-Pacheco, Juan Luna-Valdez, Antonio Luna-Figueroa, Mario Chauca, Juan Carlos Lázaro-Guillermo, et al.

    Two assumptions dominate deep learning research on pneumonia detection from chest radiographs: that newer architectures outperform older ones, and that intra-dataset evaluation suffices to establish clinical utility. This study examined both on the pediatric Chest X-Ray Images dataset. Nine architectures, spanning classical, efficient-scaling, and modern convolutional networks and Vision Transformers, plus a weighted soft-voting ensemble, were compared under a unified protocol. No post-2020 architecture provided a detectable advantage: InceptionV3 (2016) achieved the best accuracy in the main experiment (98.18%), the order of the top five models changed in an independent retraining, and Bonferroni-corrected McNemar tests found no significant differences among them. With components and weights selected on validation data, the ensemble reached 97.95% accuracy (ROC-AUC 0.997), versus 97.84% for its best component. An external evaluation on an adult RSNA subset, which combines age, institutional, acquisition, and label shifts, revealed markedly different behavior: external accuracy ranged from 71.30% to 86.20%, and VGG16, among the most accurate models internally, was the least robust. Calibration, ablation, quantitative Grad-CAM, and efficiency analyses complemented these findings. In this retrospective setting, architectural recency did not predict performance and intra-dataset accuracy did not anticipate cross-population robustness; external evaluation on independent cohorts remains necessary before clinical use.

  • Open Access

    A Hybrid Agentic RAG Architecture with Knowledge Graphs and Text-to-SQL for Automating Citizen Services in a Public Scientific Institution

    by Hugo Vega-Huerta, Leonardo Olaechea-Saavedra, Jorge Pantoja-Collantes, Gustavo Arredondo-Castillo, William Enriquez, Oscar Benito-Pacheco, Juan Honorato Luna-Valdez, Antonio Mambret Luna Figueroa, Mario Chauca, Juan Carlos Lázaro-Guillermo, et al.

    Large language models (LLMs) can produce plausible but incorrect content, a risk that is unacceptable when public institutions disclose scientific and legal information. This study presents and evaluates a hybrid agentic retrieval-augmented generation (RAG) system that automates first-level responses to citizen information requests at the Marine Institute of Peru (IMARPE). A baseline analysis of 196 requests showed a mean response time of 12 business days, above the 10-day legal limit, and identified 26.9% of them as routine, data-extractable queries. The system, developed under CRISP-DM, uses LangGraph to route each query to a LightRAG knowledge graph (Neo4j) built from institutional yearbooks, to a Text-to-SQL agent that queries the IMARSIS relational database through the Model Context Protocol (MCP), or to both, with Gemini 2.5 Flash as the foundation model and a referral protocol for out-of-scope requests. Five institutional domain experts rated 30 test cases, with mean scores of 4.45/5 for factual accuracy, 4.10/5 for completeness, and 4.57/5 for clarity. Mean end-to-end software latency was 20.43 s (maximum 49.20 s), and 30 pilot users rated usability at 80.42 on the System Usability Scale (Cronbach’s α = 0.96). These preliminary, single-site results indicate that combining graph-based and SQL-based grounding is feasible for routine citizen-information services; comparison with simpler baselines remains pending.

  • Open Access

    Deep Learning Architecture Comparison for Pneumonia Detection in Pediatric Chest X-rays: Ensemble Learning and Cross-Population External Evaluation

    by Jeanfranco Javier-Tarapaqui, Hugo Vega-Huerta, Luis Guerra-Grados, Jorge Pantoja-Collantes, Rubén Gil-Calvo, Oscar Benito-Pacheco, Juan Honorato Luna-Valdez, Antonio Mambret Luna Figueroa, Mario Chauca, Juan Carlos Lázaro-Guillermo, et al.

    Two assumptions dominate deep learning research on pneumonia detection from chest radiographs: that newer architectures outperform older ones, and that intra-dataset evaluation suffices to establish clinical utility. This study tested both. Nine architectures, spanning classical, efficient-scaling, and modern convolutional networks and Vision Transformers, plus a weighted soft-voting ensemble, were compared under a unified protocol on the pediatric Chest X-Ray Images dataset. None of the post-2020 architectures outperformed InceptionV3 (2016), which achieved the best individual accuracy (98.18%). The validation-weighted ensemble reached 98.52% accuracy and 99.22% pneumonia sensitivity on the internal test set; because its components were selected using test-set rankings, these values are upper-bound estimates. Bonferroni-corrected McNemar tests found no significant differences among the top-performing models. A cross-population external evaluation on an adult RSNA subset, which combines age, institutional, acquisition, and label shifts, revealed markedly different behavior: InceptionV3 retained 85.30% accuracy, whereas VGG16 fell to 59.60%, and the ensemble underperformed InceptionV3 alone. Ablation, Grad-CAM, and efficiency analyses complemented these findings. Architectural recency is not a proxy for performance, and external evaluation should be an architecture-selection criterion rather than an optional addendum.

  • Open Access

    Desafíos de la ingeniería de requerimientos en la metodología programación extrema (XP): una revisión sistemática

    by Oscar Benito Pacheco, Juan Luna Valdez, Edinson Montoro Alegre

    The practice of Requirements Engineering (RI) in agile development is substantially different from traditional development. Extreme programming – (XP)(Beck, 2004) is one of the approaches of these agile methods. Traditional methodologies with their waterfall proposal, in pre-established phases, are being replaced by agile methodologies, their interactive and incremental development, with nonstatic requirements. The objective of the study describes an inductive and empirical investigation on challenges and practices with the XP and IR methodology based on an analysis of data collected in academic and scientific literature. To achieve the results, a strategy was implemented that allowed document findings, following the guidelines of a Systematic Literature Review (RSL). In this way, we contribute with a description of challenges in IR in relation to the development of projects with the agile XP methodology.


  • Open Access

    Programación lineal: convergencia del algoritmo primal-dual de paso largo por el método de la función barrera

    by Juan Luna Valdez, Edinson Montoro Alegre

    En el presente trabajo se tratará de desarrollar y describir el método punto interior primal-dual para resolver el problema de programación lineal. Este método se caracteriza por utilizar funciones barrera, para el problema primal y para el dual y asi deducir el sistema no lineal primal-dual, cuya solución define la trayectoria central del método de punto interior. Se demuestra que el número total de iteraciones que ejecuta es de orden polinomial.

  • Open Access

    Aplicación del Método Espectral a la Ecuación de Aguas Poco Profundas

    by Juan Luna Valdez, Edith Carhuapoma Lopez

    En este trabajo se estudia el sistema de ecuaciones de aguas poco profundas en la forma Lagrangiana y se obtiene su solución analítica. Aplicaremos el método espectral en el análisis numérico de este sistema y se muestra que la propagación de las ondas de las aguas poco profundas no dependen del medio en que se propaga.

  • Open Access

    Solución Numérica de Ecuación de Burgers

    by Juan Luna Valdez, Edith Carhuapoma Lopez

    En este trabajo se estudia el problema de Burgers en el caso simple, se indican algunos resultados de existencia, unicidad y regularidad de la solución del problema, y en su forma variacional se da un resultado que da una condición suficiente para que una función sea solución débil. Pero como la solución puede presentar discontinuidades, entonces pasaremos a una nueva ecuación agregando un término de viscosidad artificial, obteniendo así una ecuación viscosa cuya solución, por un resultado, converge a la solución de la ecuación de Burgers cuando el término viscoso tiende a cero. Por tanto estudiaremos la ecuación viscosa desde el punto de vista numérico.

  • Open Access

    Algoritmos paralelos para la ecuación de la dinámica de fluidos

    by Juan Luna Valdez, Zaida Sales Risco

    En este trabajo se pretende desarrollar algoritmos para la simulación de problemas de la dinámica de fluidos cuyo modelo matemático esta representado por ecuaciones en derivadas parciales. Se ha tomado en cuenta que el medio es continuo donde el análisis de las partículas elementales se realiza de manera local, es decir, se busca la descripción a través de su expresión más elemental, donde las condiciones de frontera suponen que el viaje de las partículas a una distancia es constante entre una y otra, cuyas dinámicas son hasta cierto punto independientes. Se desarrolla el modelo a partir de las ecuaciones de fluidos viscosos con características homogéneas de Burgers.

  • Open Access

    Análisis de vigas por elementos finitos

    by Cristian Loli Prudencio, Edwin Chávez Ramírez, Victoriano Yauri Luque, Andrés Guardia Cayo, Amelia Villanueva Yaya, Juan Luna Valdez

    Este trabajo presenta el análisis de deformación de las vigas a través del modelo definido en el intervalo...