Isabela State University

The Cagayan River Basin (CRB), the largest river basin in the Philippines, is highly exposed to extreme rainfall and recurrent flooding. Despite this, basin-scale quantitative tools that can support both reservoir inflow forecasting and downstream flood assessment remain limited. This study assessed the use of the Rainfall-Runoff-Inundation (RRI) model to evaluate Magat Dam inflow, downstream water level, and flood inundation during major rainfall events in the CRB. The model was calibrated and tested using observed dam inflow and river water-level data from extreme events. It reproduced key hydrologic responses, particularly peak discharge and peak timing. During Typhoon Ulysses (Vamco), the model showed satisfactory performance for dam inflow (RSR 0.36, NSE 0.87, PBIAS 6.90, R² 0.88) and river water level (RSR 0.50, NSE 0.75, PBIAS − 0.39, R² 0.75). Validation during Typhoon Tisoy (Kammuri) and the December 2020 northeast monsoon further showed that the model reproduced the broad spatial pattern of observed flood depths in riverine communities, with inundation estimates ranging from 0 to 6 m. During Typhoon Ulysses, subwatershed analysis showed that Cagayan Segment 1 contributed the largest simulated peak discharge to Buntun Bridge, while downstream flooding reflected cumulative inflows from multiple upstream subwatersheds rather than Magat inflow alone. Model performance was stronger for high-flow events than for low-flow conditions, suggesting that the approach is more suitable for extreme-event forecasting than for baseflow representation. These results indicate that the calibrated RRI model can support flood early warning and dam operation in the CRB.

This qualitative case study examined the kakanin-making tradition of Cabatuan, Isabela, aimed at comprehending how local food practices become a source of cultural heritage conservation and local gastronomic tourism growth. Data for the research came from semi-structured interviews, participant observation, and document analysis of twenty (20) individuals who were purposively selected: kakanin producers, municipal officers, and long-term residents. Thematic analysis revealed five major themes: (1) kakanin-making as a living culinary heritage, (2) high community awareness and identity linkage, (3) rich but underdeveloped gastronomic tourism potential, (4) continuous challenges of modernization, marketability, and lack of institutional support, and (5) community resilience as a basis for cultural continuity and sustainable development. The findings show that the preparation of kakanin is still at the heart of Cabatuan's cultural identity. Residents have high awareness and are emotionally attached to the culture despite less participation of the youth. The study emphasizes the need to incorporate food heritage in the local tourism policy and educational programs as a means of cultural perpetuation. It ends with the statement that gastronomic culture in Cabatuan serves as a cultural narrative-a complex network of tradition, innovation, and identity that sustains rural development which is environmentally, socially, and economically viable. By maintaining its traditional foodways, Cabatuan may be able to present itself as a heritage-based gastronomic tourism prototype not only to the rest of the country but also to the world, the Philippines.

This study investigates the physicochemical and sensory factors influencing the quality of cacao wine–based beverages using machine learning techniques. A dual-dataset approach was employed, integrating a secondary wine quality dataset containing physicochemical properties and a primary dataset derived from sensory evaluation of cacao wine–infused cocktails. The primary dataset included attributes such as color, taste, fineness, aroma, alcohol content, and overall acceptability. A Random Forest Regression model was applied to both datasets to evaluate predictive performance and identify key determinants of quality. The wine dataset yielded a moderate predictive performance with an R2 score of 0.551, Mean Squared Error (MSE) of 0.348, and Root Mean Squared Error (RMSE) of 0.5897, with alcohol, volatile acidity, and free sulfur dioxide identified as the most influential variables. In contrast, the cacao dataset demonstrated strong predictive performance, achieving an R2 score of 0.7895, MSE of 0.035, and RMSE of 0.1873. Feature importance analysis revealed that fineness (0.28) and aroma (0.28) were the most significant predictors of overall acceptability, followed by taste (0.23) and color (0.21), while alcohol content (0.01) had minimal influence. The findings highlight that while physicochemical properties contribute to the formation of beverage characteristics, sensory attributes play a more dominant role in determining consumer acceptability. The study demonstrates the effectiveness of machine learning in identifying key quality drivers and provides a data-driven framework for optimizing cacao wine–based beverages.

This study presents a Multi-Fusion Multi-Staged Transfer Learning framework integrated with Contrast Limited Adaptive Histogram Equalization (CLAHE) for the classification of lung disease using chest X-ray images. A total of 4,862 images obtained from Kaggle were utilized and divided into training (3,403), validation (729), and testing (730) subsets. CLAHE preprocessing was applied to enhance image contrast and improve feature visibility prior to model training. The proposed framework combines MobileNetV2 and EfficientNetB4 through feature fusion and employs a three-stage fine-tuning strategy to progressively adapt the model to domain-specific features. Experimental results show that the proposed model achieved a test accuracy of 93.42% with a corresponding loss of 0.1694. Class-level evaluation indicates a precision of 0.9909 for tuberculosis detection and a recall of 0.9983 for normal classification, demonstrating strong reliability and sensitivity. The integration of CLAHE, feature fusion, and multi-stage training significantly improved model performance and convergence stability. The findings suggest that the proposed framework is effective for lung disease classification and has potential applications in computer-aided diagnostic systems, particularly in supporting rapid and accurate screening. The novelty of this work lies in the unified combination of image enhancement, dual-backbone fusion, and progressive transfer learning within a single framework.

This study aimed to explore the Indigenous Identity ‘Tayo” of IP youth in San Mateo, as a basis for developing localized instructional materials in teaching Understanding Culture, Society, and Politics (UCSP). The study used qualitative phenomenology research design to understand the lived experiences of IP youth in relation to their cultural practices, values, and identity formation, as well as how these are reflected in existing school instructional materials. Data were collected through interviews with six (6) indigenous youth participants. The responses were analyzed using thematic analysis, findings revealed six major themes such as cultural practices, core values, social influences, and gaps in instructional materials. It also indicates that while the “Tayo” identity remains deeply rooted among IP youth, it is continuously evolving in response to modern influences. Moreover, existing instructional materials were found to inadequately represent Indigenous identity, highlighting the need for culturally responsive and localized content. Based on these findings, the study proposes a conceptual framework to guide the development of instructional materials that integrate Indigenous perspectives into the UCSP curriculum. The study concludes that integrating the lived experiences and cultural realities of Indigenous youth into educational materials can enhance relevance, inclusivity, and cultural sensitivity in teaching. It is recommended that educators, curriculum developers, and policymakers should prioritize localization and Indigenous representation in instructional design.