Vitervo López Caballero

The effects of climate change have altered precipitation patterns, causing extreme weather events that impact crop yields and increase the demand for agricultural production, which drives the need to implement emerging technologies. This research presents the design, construction, and validation of a real-time monitoring system for environmental parameters in open-field avocado orchards. A four-phase methodology was employed: (1) Establishment of required parameters and construction of the IoT architecture; (2) Design of the geometry and final elaboration of the system: modeling in SolidWorks, circuit diagrams in Fritzing, and system assembly; (3) Development of a mobile application: Android Studio and development of the autoencoder model with machine learning; and (4) Validation: Evaluation of the wireless link; system implementation and deployment of the mobile application. The results obtained include a system with a transmitter node equipped with sensors and a receiver, both incorporating ESP32 and nRF24L01+PA+LNA modules, with a wireless transmission range of 2300 m. A total of 2016 data records were stored on microSD and in the cloud, which can be queried, visualized, and analyzed through the mobile application via Bluetooth or Wi-Fi; the system also detects outlier data that can be used for decision-making in the agricultural sector.

Diabetic foot ulcers represent a severe complication of diabetes mellitus, affecting millions of adults worldwide and often leading to hospitalization and amputation. Diabetic neuropathy increases the risk of plantar injuries, while the lack of continuous monitoring and delayed detection contributes to the progression of these lesions. This study presents a low-cost smart insole system for continuous plantar pressure monitoring and screening of plantar pressure patterns associated with diabetic neuropathy. The system integrates piezoresistive sensors distributed across key regions of the foot, connected to a low-power ESP32 microcontroller for data acquisition. Measurements are transmitted via Bluetooth Low Energy to a mobile application that enables real-time visualization, user management, and storage in a MySQL database for historical data consultation. Data processing employs a convolutional neural network configured to classify plantar pressure patterns between non-diabetic individuals and diabetic patients presenting neuropathic alterations. System validation demonstrated 88% accuracy, 88% recall, and 87% F1-score in classifying plantar pressure patterns. The results confirm that the combination of low-cost hardware and open-source software constitutes a viable and scalable solution for screening biomechanical alterations associated with diabetic foot complications.

Poverty in Mexico is measured biennially through socioeconomic surveys identifying people with income deficiencies and six social dimensions. This approach faces significant limitations: high costs, operational difficulties in regions with violence, and restricted capacity for timely policy responses, as evidenced during COVID-19. Emerging methods based on satellite imagery and remote sensing expand poverty analysis scope and efficiency. However, implementation faces challenges from scarce primary data, particularly satellite imagery and field measurements essential for predictive model training. This article develops a multispectral database for Mexico designed to facilitate training and validation of advanced machine learning algorithms for poverty prediction. This contribution provides a methodological tool for designing robust and scalable approaches while reducing reliance on traditional socio-economic survey methods. Spanish-language metadata / Metadatos en españolTítulo en español: Creación de un conjunto de datos multiespectrales a partir de imágenes satelitales para la predicción de la pobreza a nivel municipal en México Resumen: En México, la pobreza se mide cada dos años mediante encuestas socioeconómicas que identifican a las personas con ingresos insuficientes y tienen en cuenta seis dimensiones sociales. Este enfoque presenta importantes limitaciones: altos costos, dificultades operativas en regiones afectadas por la violencia y una capacidad limitada para dar respuestas políticas oportunas, como quedó demostrado durante la pandemia de COVID-19. Los métodos emergentes basados en imágenes satelitales y teledetección amplían el alcance y la eficacia del análisis de la pobreza. Sin embargo, su implementación se enfrenta a dificultades derivadas de la escasez de datos primarios, en particular de imágenes satelitales y mediciones sobre el terreno, que son esenciales para el entrenamiento de los modelos predictivos. Este artículo presenta una base de datos multiespectral para México diseñada para facilitar el entrenamiento y la validación de algoritmos avanzados de aprendizaje automático para la predicción de la pobreza. Esta contribución ofrece una herramienta metodológica para diseñar enfoques robustos y escalables, al tiempo que reduce la dependencia de los métodos tradicionales de encuestas socioeconómicas. Palabras Claves: Conjunto de datos, pobreza, imágenes satelitales, aprendizaje automático Smart citations: https://scite.ai/reports/10.61467/2007.1558.2026.v17i3.1283Dimensions.Open Alex.

La Minería de Datos Educativos (EDM) aplica métodos de ciencia de datos para analizar información académica y apoyar la toma de decisiones basada en evidencia. Este artículo de investigación, realizado en una Institución de Educación Superior en México, identifica patrones asociados con éxito o riesgo académico en estudiantes de licenciatura mediante cuatro fases: recopilación, depuración, modelado y análisis. Con técnicas no supervisadas como PCA y K-means se detectaron tres perfiles de estudiantes, uno de ellos con señales de desinterés emocional o académico. En la etapa predictiva, modelos supervisados como MLP, SVM y Regresión Logística alcanzaron precisiones del 95–96%. Los resultados muestran que el aprendizaje automático es clave para identificar estudiantes en riesgo y mejorar la intervención institucional.

El propósito de este trabajo es presentar una revisión del estado del arte sobre los algoritmos más empleados en la medición de la pobreza a nivel global, con el fin de identificar enfoques analíticos vigentes y ofrecer estrategias que contribuyan a abordar este problema complejo y dinámico. La revisión realizada evidencia que los algoritmos con mayor presencia en la literatura reciente son las Redes Neuronales Convolucionales (CNN), Random Forest (RF), Gradient Boosting (incluido XGBoost) y las Redes Neuronales no Convolucionales. Las Redes Neuronales Convolucionales sobresalen por su alta efectividad en el procesamiento y extracción de características visuales, mientras que Random Forest y Gradient Boosting destacan por su versatilidad, capacidad predictiva y robustez en contextos diversos.

The article was not new and interesting to me. As I was reading it, I thought at one point that I was reading a Cloud Computing blog. The article lacks experimental results. They never mention where they test the proposed model. On the other hand, th...

E-learning resources are pieces of knowledge used for achieving a learning objective. One of the problems in e-training in the workplace has to do with the lack of relevance of learning resources. Here, we define a learning resource as a resource that contains information about a task or refers to a task realized within the context of an organizational process. We describe a set of activities for the generation of learning resources of objective, content, activity, and assessment types, which correspond to the pedagogical elements in learning objects. The generation goes from a documented business process using natural language processing and the learning resources produced are packaged as web services for independence and reusability purposes. Each learning resource generated is independent since it may participate in several formative sequences on its own, and it is reusable since it may be used many times without any change.

The K-means clustering algorithm is widely used in several domains, because of its simplicity of implementation and interpretation. However, one of its limitations is its high computational complexity. In this work the problem of reducing the complexity of the K means algorithm is approached, in order to make possible the solution of large scale data sets like those from Big Data, without significantly degrading solution quality. To this end, a new metaheuristics is proposed, which by an early assignment of objects to clusters, significantly reduces the number of calculations of distances from objects to centroids. The approach was experimentally evaluated by solving real and synthetic datasets yielding encouraging results. Time reductions of up to 91% were obtained with respect to the standard K-means, at the expense of reducing quality by 3.2%.