Subclinical atherosclerosis (SCA) has been associated with lung function in many populations. Accordingly, those factors involved in respiratory health may contribute to the mechanisms linking pulmonary and cardiovascular disorders. Since the airway microbiome plays a key role in respiratory pathophysiology, microbial abundance and diversity in the upper airway may be associated with SCA risk. This study aimed to assess the association of upper airway microbiome abundance and diversity with SCA risk. A nested case–control study was carried out among adults from the Southern Caribbean. SCA was defined by carotid Doppler ultrasound as a carotid intima-media thickness ≥ 0.8 mm or the presence of carotid plaque. Oropharyngeal mucosal samples were collected under fasting conditions, and the V3-V4 region of the 16SRNA gene was sequenced. Differences in genus-level abundance were assessed using Microbiome Multivariable Association with Linear Models 2 (MaAsLin2). Alpha-diversity indices were compared through Wilcoxon tests. Principal coordinates analysis (PCoA) and Permutational Multivariate Analysis of Variance (PERMANOVA), adjusted for age, body mass index, and smoking status, were performed to assess associations between Bray–Curtis distances and SCA. MaAsLin2 was also applied to determine predicted metabolic pathway enrichment. A total of 40 cases and 40 controls were analyzed. The oropharyngeal microbiomes composition was dominated by three genera: Prevotella (cases: 26.0%; controls: 29.8%), Veillonella (cases: 16.3%; controls: 13.7%) and Fusobacterium (cases: 7.5%; controls: 8.0%). Significant differences in relative abundance were found for Streptococcus, Aureimonas, Mogibaecterium, Candidatus Saccharimonas, Pseudomonas, Segatella, Stomatobaculum, Oribacterium and Lachnoanaerobaculum, with lower abundance in cases for all these genera. Alpha-diversity was lower in cases than in controls, as reflected by the Shannon index (cases: 5.95 IQR [4.87; 6.35] vs. controls: 6.49 IQR [6.29; 6.75], p < 0.001), Simpson index (cases: 0.994, IQR [0.989; 0.996] vs. controls: 0.997 IQR [0.997; 0.998], p < 0.001) and Chao1 richness (cases: 1076.2 IQR [345.2; 1473.4] vs. controls: 1136.1, IQR [841.1; 1635.3], p = 0.04) were compared. In adjusted-PERMANOVA, 8.2% of microbiome variations were associated with SCA (pseudo-F = 7.49, R2 = 0.082, p < 0.001), although this finding may have been influenced by intra-group dispersion among cases (PERMDISP: pseudo-F = 5.62, p = 0.018; Tukey’s HSD test: mean difference = −0.071, 95% CI [−0.118, −0.025], p = 0.003) Predicted enrichment of phenylalanine metabolism and indole alkaloid biosynthesis was higher in cases. In conclusion, these exploratory analyses suggest that reduced oropharyngeal microbiome diversity is associated with SCA. Further studies are warranted to clarify the role of the airway microbiome on cardiovascular outcomes.
Automated plant disease recognition is increasingly used as a benchmark application for computer vision in precision agriculture, yet the relative benefit of classical machine learning and transfer learning depends on feature representation and experimental design. This study presents a reproducible comparison of K-Nearest Neighbors (KNN), Support Vector Machine (SVM), Random Forest (RF), and MobileNetV2 for ten-class tomato leaf classification using the PlantVillage dataset. The 18,178 original images were partitioned class-wise into training (12,720), validation (2,722), and independent test (2,736) subsets. Only the training subset was balanced through image augmentation, producing 37,490 training images. Classical models used a 110-dimensional descriptor combining HSV color histograms, gray-level co-occurrence matrix statistics, and local binary patterns, whereas MobileNetV2 used ImageNet transfer learning with model-specific preprocessing and two-stage fine-tuning. On the independent test set, MobileNetV2 achieved the highest numerical accuracy (97.26%), SVM-RBF achieved 97.11% accuracy and the highest macro-F1 score (96.47%), and RF achieved 96.93% accuracy and the highest macro-recall (96.69%). KNN achieved 58.22% accuracy. Wilson confidence intervals and stratified bootstrap analysis showed closely overlapping uncertainty ranges for SVM-RBF, RF, and MobileNetV2. Exact paired McNemar tests with Holm correction found no statistically significant differences in classification correctness among these three high-performing models, whereas each significantly outperformed KNN. The results indicate that compact handcrafted color-texture representations combined with nonlinear classical classifiers can achieve performance comparable to transfer learning under controlled PlantVillage conditions. External validation with field-acquired images remains necessary before inferring real-world diagnostic performance.
La gestión de los proyectos de extensión universitaria en Colombia requiere una transformación funcional que supere los esquemas tradicionales de transferencia unidireccional y prestación de servicios. Este artículo presenta el diseño y validación de un modelo de innovación social estructurado para articular las capacidades de las instituciones de educación superior (IES) con las demandas y saberes de los territorios. Mediante un enfoque metodológico mixto, de alcance descriptivo-correlacional, se encuestó y analizó una muestra representativa de 56 universidades colombianas, integrando además el estudio profundo de tres casos institucionales de éxito. Los resultados demuestran que la integración de cinco dimensiones estratégicas —co-creación comunitaria, gobernanza flexible, pertinencia territorial, impacto sistémico y sostenibilidad— explica el 83.16% de la varianza del desempeño extensionista (p < 0.05). Se concluye que el modelo propuesto incrementa la pertinencia institucional y optimiza la solución colaborativa de problemáticas complejas.
Introduction : Giardia intestinalis (G. intestinalis) is an intestinal parasite that infected between 200 and 300 million people worldwide. Some patients developed reactive arthritis after infection clearance. Molecular mimicry could induce cross reactivity to generate autoimmune disease. Objective : To evaluate the potential molecular mimicry between autoantigens implicated in in reactive arthritis and G. intestinalis antigens using in silico analysis. Methodology : Similarity between autoantigens and protozoa antigens was analyzed using PSI-BLAST. After, shared regions were identified performing binary alignments. Using 3D structures retrieved from protein data bank or modelled B cell epitopes were predicted with Ellipro tool. Epitopes predicted were visualized on 3D structure with Pymol software. Results : In total, six of thirty-nine autoantigens report for reactive arthritis-associated showed similarity with proteome reported for G. intestinalis. autoantigens were included in the study and compared against the G. intestinalis proteome. These autoantigens shared up 30% with homologues from protozoa. We have that the autoantigens that showed the greatest identity were Alpha enolase with Enolase and HSP 60 Chaperonin 60, Putative TCP-1/cpn60 chaperonin family protein, 60 kDa of groel chaperonin. Sixty-eight epitopes with scores greater than 0.7 were identified from G. intestinalis antigens. Conclusion : Six G. intestinalis antigens are identified that mimic reactive arthritis autoantigens, with moderate levels of identity that could explain a cross-reactivity involved in the development of the autoimmune response reported in patients who suffered from the infection.
El presente artículo presenta un estado del arte de la investigación científica reciente en el campo de la educación, la pedagogía y la didáctica, con énfasis en los aportes teóricos y empíricos sobre la diversidad y la inclusión. A partir de una revisión sistemática de literatura fundamentada en lineamientos PRISMA y de artículos indexados en bases de datos de alto impacto como Scopus, se examinan las principales líneas de producción académica entre 2021 y 2026. Los resultados evidencian un tránsito conceptual que desplaza la mirada desde la integración de poblaciones con necesidades especiales hacia la transformación estructural de los sistemas escolares. Se identifican dimensiones críticas como las prácticas pedagógicas inclusivas, las culturas institucionales y la formación docente, concluyendo que la diversidad no debe asumirse como un déficit por corregir, sino como un eje estructural de la justicia social y el derecho universal a la educación.