Scopus 2025
Permanent URI for this collectionhttps://repository.li.mahidol.ac.th/handle/123456789/102712
Browse
Recent Submissions
Item Metadata only Effects of Caffeine on Kidney Stone Disease and Renal Cancer(2025-05-23) Peerapen P.; Thongboonkerd V.; Peerapen P.; Mahidol UniversityCaffeine is a natural xanthine alkaloid commonly found in coffee beans, cocoa beans, and tea leaves. It acts as a non-selective adenosine receptor antagonist and can block adenosine receptors expressed on renal tubular epithelial cells, leading to diuresis and natriuresis. Caffeine has long been known to have beneficial effects on human health, as it can reduce risks of many diseases. Several lines of recent evidence have shown the protective roles of coffee and caffeine consumption in kidney stone disease and renal cancer. This chapter thus provides an overview of the physiological roles of caffeine in kidney functions and summarizes the previous and recent data on the roles of coffee and caffeine consumption in kidney stone disease and renal cancer.Item Metadata only Domain-Specific Named Entity Recognition in Hotel Reviews Using Large Language Models(2025-01-01) Tiebrat K.; Chuckpaiwong R.; Samanchuen T.; Tiebrat K.; Mahidol UniversityExtracting meaningful insights from large volumes of hotel review text remains a key challenge in natural language processing. Traditional methods often rely on manual feature design or limited rule-based systems. This study introduces an automated workflow that uses large language models (LLMs) to transform unstructured hotel reviews into structured, analytical data. The process integrates predefined prompts with the review texts to guide the model in identifying entities, classifying them into relevant aspects, and determining sentiment polarity. The approach effectively captures contextual information and produces highly interpretable, structured outputs. By leveraging prompt engineering and LLM inference, it reduces human intervention while maintaining analytical depth. The proposed workflow demonstrates the potential of LLM-driven automation to streamline sentiment and aspect extraction from usergenerated content, enabling scalable and data-driven insights in hospitality and related domains.Item Metadata only Improving Course Registration Efficiency Through LINE Chatbot with LLM Integration(2025-01-01) On-Uean K.; Chaikot A.; Samanchuen T.; Pookkaman W.; On-Uean K.; Mahidol UniversityThis study addresses inefficiencies in course registration at a private training center that used Google Forms. The process resulted in redundant data entry, a 25% error rate, and no real-time status tracking. The study pursued three objectives 1) developing an Automated Question Answering System (AQAS) using LLM technology 2) implementing an Automated Course Registration System (ACRS) via LINE Messaging API and 3) minimizing data entry errors to enhance data management efficiency. System development followed the Software Development Life Cycle (SDLC) waterfall model, the system was developed through structured phases of analysis, design, implementation, testing, deployment, and maintenance. Data management utilizes MySQL database with administrator validation conducted directly through LINE. System evaluation presented improvements in all three objectives. The LLMintegrated question-answering system achieved 86% accuracy across 50 structured queries. Comparative testing with 80 participants across three courses showed that the LINE-based registration system improved registration data entry accuracy by a 26.7% relative increase and reduced processing time by 50-66 %. These improvements enhanced user interaction quality, reduced staff workload, and demonstrated the effectiveness of AI-enabled chatbots for educational administration.Item Metadata only Toxic Leadership Influencing Turnover Intention of ICT Staff in the University(2025-01-01) Fongkum N.; Kirivan V.; Leelasantitham A.; Fongkum N.; Mahidol UniversityHigher education institutions are increasingly experiencing staff turnover, particularly among information and communication technology (ICT) personnel, which poses challenges to operational efficiency and continuity. This study examines the impact of toxic leadership on ICT staff turnover intention. It further investigates how factors such as satisfaction of self-determination, workplace as a learning environment, growth needs, work engagement, and emotional exhaustion are related to turnover intention. The findings offer insights into how leadership and organizational factors influence employee retention and provide guidance for strategies aimed at improving workplace well-being.Item Metadata only Design of an Agentic-AI-Based Accessible Library Service for Persons with Disabilities(2025-01-01) Tonnondaeng P.; Maneechay P.; Warinsiriruk E.; Wang Y.T.; Tonnondaeng P.; Mahidol UniversityTo address persistent barriers in information access for persons with visual impairments, this paper proposes a conceptual design of an Agentic-AI-enabled accessible library service model that supports more independent use of library resources. The system integrates two core components: an AI Agent that answers general library inquiries and book-location queries via natural-language voice interaction, and an Optical Character Recognition (OCR) based reading module that converts printed text into digital text and delivers it as synthesized speech. Orchestrated through automated workflow, these components form an autonomous, staff-independent service workflow that reduces reliance on human assistance. System performance was evaluated in two experimental modules: Module 1 assessed the AI Agent's ability to respond to typical librarian-user queries, and Module 2 evaluated the accuracy of searching for and identifying shelf locations of specific books, while the OCR module was tested on Thai text, English text, and mathematical expressions. The results showed 0 % residual error in reconstructing Thai and English descriptive text, whereas mathematical content remained challenging. Overall, the findings support the technical feasibility of the proposed model as a foundation for more inclusive, AIsupported library services for visually impaired users.Item Metadata only Enterprise Architecture: Myth, Reality, and Success Factors(2025-01-01) Mayakul T.; Hiranchan S.; Mayakul T.; Mahidol UniversityEnterprise Architecture (EA) has evolved into a critical strategic framework for digital transformation, yet implementation failure rates remain consistently high across both government and private sector organizations. A significant gap persists between theoretical EA benefits and practical implementation outcomes, necessitating systematic investigation of success and failure factors. The system reviews revealed six core success factors: strong governance and leadership, effective communication and collaboration, organizational integration, skilled teams, strategic alignment, and appropriate methodologies. Communication emerged as the critical enabler for change management, while organizational anchoring mediated other factors' effectiveness. Government organizations emphasized governance structures and regulatory compliance, whereas private enterprises focused on agility and return on investment. Both sectors required continuous evaluation mechanisms for sustained EA benefits. The findings conclusively demonstrate that EA success depends on sustained organizational work rather than framework selection or technological solutions alone. Effective implementation requires establishing clear to-be vision, comprehensive planning across business-data-application-technology domains, and ongoing commitment to architectural discipline. This research bridges the theory-practice gap by providing evidence-based guidance for organizations seeking EA transformation success. Enterprise Architecture works only when organizations commit to doing the necessary work.Item Metadata only Smart Gas Cylinder Warehouse Management Using Computer Vision for Automated Inventory, Classification, and Safety Monitoring(2025-01-01) Sa-Nga-Ngam P.; Athikulrat K.; Pakpoom P.; Sa-Nga-Ngam P.; Mahidol UniversityThis research applies the YOLOv8 model, running on local edge computing, for real-time gas cylinder detection and classification. The system was trained to identify and distinguish gas types (classification) and to analyze the physical state of the cylinders (posture analysis) in two forms that can lead to warehouse hazards or accidents: tilted and fallen. The technical performance evaluation revealed that the developed model achieved an mAP@ 0.5 of 93.8% for overall gas cylinder detection. Moreover, its safety monitoring performance demonstrated a recall rate of 99.0% for detecting fallen/tilted cylinders. This system is integrated with the WMS (warehouse management system) via an API service running in a cloud container to automatically update cylinder locations and stock counts. It also provides alerts if cylinders are placed in the wrong zone. The results demonstrate the potential to reduce stock-taking time by 4 5. 5 5 %, increase inventory accuracy by 4 9. 0 1 %, and sustainably elevate safety standards within the warehouse. These contributions support the advancement of industrial innovation and smart infrastructure (SDG 9) while enhancing workplace safety and operational efficiency (SDG 8).Item Metadata only Comparative Evaluation of Large Language Models for Automated Text Classification and Data Labeling in Drug Recommendation(2025-01-01) Rattanachayabun K.; Jamrat S.; Samanchuen T.; Rattanachayabun K.; Mahidol UniversityData labeling remains one of the most laborintensive stages in artificial intelligence (AI) development, particularly in natural language processing. This study investigates the potential of large language models (LLMs) to automate text classification and reduce the burden of manual annotation. Four state-of-the-art LLMs-GPT-4o, Llama 3.3, Llama 3.1, and Gemma 2-were evaluated on a drug recommendation dataset using three prompting strategies: zero-shot, one-shot, and few-shot learning. Model performance was assessed using F1-scores across therapeutic consideration, use as directed, and contraindication categories. The results indicate that incorporating even a single labeled example substantially improves classification accuracy compared with zero-shot prompting, while the performance gain from one-shot to few-shot prompting is marginal. Among the tested models, Llama 3.3 achieved the most consistent results, whereas Gemma 2 demonstrated strong zeroshot generalization and GPT-4o provided stable cross-strategy performance. The findings highlight the feasibility of employing LLMs for automated data labeling and underscore the efficiency of one-shot prompting as a practical balance between accuracy and computational cost.Item Metadata only Evaluating and Comparing Machine Learning Models for PM2.5 Health Impact Assessment in Thailand(2025-01-01) Nateeprasittipon P.; Sa-Nga-Ngam P.; Chansutthirangkool M.; Nateeprasittipon P.; Mahidol UniversityThis study presents a methodological comparison of four machine learning and statistical models to assess the acute impact of fine particulate matter (PM2.5) on public health in Northern Thailand. The performance of Random Forest (RF) and Long Short-Term Memory (LSTM) was evaluated against two baselines: Linear Regression (LR) and SARIMAX. Air quality data and health surveillance data from 2023 were integrated for four disease groups in Health Region 2. The final merged datasets comprised approximately 5 0 0 - 6 0 0 provinceday records per disease group. The results demonstrate that Random Forest was the most robust model, achieving the highest R -squared scores for Skin (R2=0.21) and Eye diseases (R2 = 0. 1 8). The LSTM model showed competitive performance, ranking second, whereas Linear Regression failed to capture the non-linear patterns. Conversely, all models yielded negative R2 values for Cardiovascular disease, suggesting that short-term exposure (2-day lag) is insufficient for predicting heart-related conditions.Item Metadata only A Business Intelligence Dashboard for Radiological Technology Education in the AUN-QA 4.0 Framework(2025-01-01) Manoi S.; Chujai L.; Leelasantitham A.; Manoi S.; Mahidol UniversityThis paper presents a Business Intelligence (BI) dashboard for the digital transformation of quality assurance (QA) in higher education, aligning with Digital Education and Healthcare Information Technology Management. We developed the dashboard to monitor ASEAN University Network-Quality Assurance (AUN-QA) 4.0 learning outcomes in a Radiological Technology program. The process involved defining AUN-QA 4.0 KPIs, integrating disparate institutional data, and deploying the interactive system. The resulting dashboard provides real-time visualization and actionable insights, enhancing transparency and facilitating data-driven decision-making. This study demonstrates BI as a scalable model for Digital Transformation in other compliance-driven academic programs, particularly within Digital Healthcare.Item Metadata only AutoGPT Devloop: An Autonomous AI Development Framework for End-to-End Software Generation, Execution, and Self-Repair(2025-01-01) Asavathongkul T.; Sa-Nga-Ngam P.; Kiattisin S.; Asavathongkul T.; Mahidol UniversityWe present the Autonomous AI Development Framework (AADF), a self-developing agent that converts high-level goals into functioning software through iterative planning, code generation, execution, and self-repair. AADF integrates (i) task decomposition, (ii) a semantic code memory (vector database), (iii) virtualized environment management, and (iv) automated file/version control to achieve autonomy across the software lifecycle. Using a Design Science Research approach, we evaluate AADF on a suite of 15 GUI / web tasks spanning five difficulty levels (3 trials each; 45 trials). The framework achieves 100% success on Levels 1-2 and Level 4, with partial degradations on tasks dependent on external API credentials or large NLP resources (Level 3: 67 % / 33 % on two tasks; Level 5: 67% on one task). Overall, AADF completes 41 / 45 trials (91.1 %) without manual code editing, demonstrating reliable self-repair for dependency conflicts and missing imports, and reducing human effort on environment setup, boilerplate, and repetitive file operations. We discuss error taxonomies (credentials, heavyweight downloads, concurrency), autonomy-speed trade-offs, and actionable design guidelines for safe deployment in practice.Item Metadata only Optimizing Optical Character Recognition Within a Physical - Agentic AI System for Flexible Drug Preparation(2025-01-01) Maneechay P.; Warinsiriruk E.; Wang Y.T.; Maneechay P.; Mahidol UniversityThe conventional camera-based prescription-label reading process used in existing automated systems has notable limitations in both accuracy and latency. These issues stem primarily from Optical Character Recognition (OCR) pipelines that were not optimized for real-world label characteristics-such as varying font complexity, size, and image quality-resulting in misread text and delays that fail to meet operational requirements. To address these shortcomings, this study developed an improved processing pipeline by comparing the performance of EasyOCR and PyTesseract under image-downscaling conditions ranging from 0.1 to 0.9. In parallel, an integrated N8N-AI Agent workflow was designed to enhance both the speed and accuracy of medication-label extraction. The proposed system combines appropriate pre-processing, selective OCR utilization, and the incorporation of reference data directly within the model. This integration leads to more stable label-reading performance, enabling the system to correctly identify medication names while reducing overall processing time compared with the previous approach. Experimental results show that PyTesseract processes images approximately 5-10 times faster than EasyOCR, whereas EasyOCR consistently delivers higher recognition accuracy. When combined with reference data, the workflow using a system-prompt approach proved more than ten times faster than the CSV-based lookup method. Optimizing the OCR for image complexity, minimizing node count, and applying in-memory processing collectively improved both the responsiveness and accuracy of the system. As a result, the new pipeline operates near real time, reduces bottlenecks associated with redundant file operations, and maintains stable performance across diverse medication-label formats-an essential requirement for reliable deployment in medical environments where precision and consistency are critical.Item Metadata only Predictions of Stem Skills in Human Resource Management Using Machine Learning(2025-01-01) Nuntasing S.; Leelasantitham A.; Sukamongkol Y.; Nuntasing S.; Mahidol UniversityGovernments are now widely adopting Information Technology (IT) as a critical tool for reforming and driving public sector organizations. STEM (Science, Technology, Engineering, and Mathematics) skills are fundamental for analyzing and executing initiatives related to this IT-driven transformation. This research focuses on applying Machine Learning (ML) techniques to human resource management. By efficiently analyzing data from HRM systems, the goal is to improve the planning of human resource development. The objective of this research is to optimize outcomes and determine the best-performing Machine Learning (ML) model. This research compared various Machine Learning models and techniques for handling imbalanced data to evaluate their predictive performance for a target group. The models considered include K-Nearest Neighbors, Random Forests, Support Vector Machines, Multi-layer Perceptron, Cat Boost, and others. These models were evaluated based on their accuracy in predicting the potential of the target group. The results indicate that K-Nearest Neighbors and Random Forests are the topperforming models, with superior accuracy and strong ROC curve performance. These findings suggest that these models are wellsuited for evaluating the target group's scores. Additionally, the performance of other models was discussed, highlighting their respective strengths and weaknesses in various scenarios. This research is of paramount importance for the planning and development of human resources, as well as for advancing organizational progress. By applying the results from Machine Learning (ML), we can enhance efficiency and strategize Human Resource Management (HRM) more effectively. Furthermore, this study underscores the necessity of selecting an appropriate ML model for specific tasks to maximize organizational benefits.Item Metadata only The Convergence of Forensics and Spatial Computing: A Comprehensive Analysis of Mixed Reality's Role in Modern Crime Scene Investigation(2025-01-01) Werukanjana P.; Malikhao R.; Sa-Nga-Ngam P.; Permpool N.; Israngkul W.; Muttitanon W.; Werukanjana P.; Mahidol UniversityThis experiment proves a Crime Scene Investigation (CSI) system that uses Mixed Reality (MR) to solve key shortcomings of traditional CSI methods, such as evidence, chain of custody, and contamination. Research and development applications are used to study emerging mixed reality technologies with high-fidelity professional teams and on-site fieldwork. The extended study improves the CSI system by streamlining evidence collection in realism assessment, reducing evidence contamination, and enabling collaborative teamwork. We evaluated this exam on usability and satisfaction. 1) The System Usability Score (SUS) was above average at 69. The high score was due to show tag and symbol marker substitution usability metrics. 2) Quantitative and qualitative findings were confirmed by the clear correlation between perceived efficiency and observed time reductions in evidence collection. User feedback technology is great, and the collaborative aspects assist them in their work. This analysis is supported by empirical case studies and concludes with a forward-looking perspective on the future integration of MR with other FBI recommendation steps and evidence, advocating for a robust framework to guide its responsible and effective use in law enforcement and the justice system.Item Metadata only Impact of low vision and blindness on characteristics of developmental delay in children younger than 6 years(2025-12-09) Wannapaschaiyong P.; Chotikavanich S.; Sutchritpongsa S.; Rojmahamonkol P.; Penphattarakul A.; Saksiriwutto P.; Eiamsamarng A.; Setthawong S.; Phongsuphan T.; Jaruniphakul P.; Yingyong R.; Sarinak N.; Eksupapan E.; Sagan S.; Onlamul P.; Wannapaschaiyong P.; Mahidol UniversityBACKGROUND Visual impairment during early childhood can hinder motor, language, and social development, yet data on its developmental impact across common pediatric oc ular diseases remain limited. AIM To investigate the developmental impact of low vision and blindness on children under six with common ocular diseases. METHODS This retrospective study reviewed records of new patients under six with visual impairment at Siriraj Hospital’s low vision rehabilitation center (January 2017 October 2022). We collected ocular, systemic, and developmental data; recorded visual acuity in the better-seeing eye after refractive correction; and assessed developmental domains with the Denver II. Univariable and multi-variable logistic regression identified factors associated with developmental delay. RESULTS A total of 161 pediatric patients (mean age 24.9 ± 18.9 months) were enrolled and evaluated based on their ability to fix on and follow an object or light source. Some were further assessed using the Allen picture chart and all had visual acuity worse than 1.07 ± 0.58 LogMAR, and 83.2% were identified as having global developmental delay (GDD). The three most common ocular causes were cortical visual impairment (CVI), optic neuropathy/atrophy, and optic nerve hypoplasia. Extremely poor visual acuity (inability to fixate and follow) was significantly associated with GDD [adjusted odds ratio (AOR) 41.0] and delays in all developmental domains: Gross motor (AOR 10.0), fine motor (AOR 12.8), language (AOR 5.3), and personal-social skills (AOR 13.4) (P ≤ 0.002). Multiple disabilities, most often visual impairment with cerebral palsy, were also significantly associated with gross motor delays (AOR 7.7) and fine motor delays (AOR 4.0) (P < 0.05). CVI was also related to delays in language and personal-social skills (AOR 9.1 each) (P < 0.05). CONCLUSION This study underscores the developmental issues in children with visual impairment, especially those with poorer acuity, CVI, and multiple disabilities. Significant delays were observed in all domains, including GDD. A timely referral to specialists is strongly recommended.Item Metadata only Tourist Migration from Russia to Thailand: Current Trends and Development Prospects(2025-01-01) Ryazantsev S.V.; Rakhmonov A.K.; Ryazantsev N.S.; Teerasenee T.; Ryazantsev S.V.; Mahidol UniversityThe article explores the trends and future prospects of Russian tourist migration to Thailand, examining the factors that have shaped travel patterns over the past decade. Thailand has consistently remained one of the most popular destinations for Russian travelers, owing to its warm climate, affordability, and welcoming tourism policies. However, fluctuations in the number of Russian tourists have been influenced by economic conditions, geopolitical factors, and global crises, particularly the COVID-19 pandemic. The research highlights the seasonal nature of Russian tourism, with peak travel occurring during the winter months as Russian tourists seek warmer climates. Policy measures, such as Thailand’s visa-free travel regime and the acceptance of Russian banking systems, have played a crucial role in attracting and maintaining Russian tourist inflows. Furthermore, Thailand’s strategic position as an alternative destination amid Western sanctions has made it an even more appealing choice for Russian travelers. Beyond travel trends, the study also underscores the significant economic impact of Russian tourism on Thailand. Russian visitors contribute substantially to the country’s hospitality, transportation, and retail sectors, supporting local businesses and employment. Despite challenges such as economic instability and environmental concerns, Thailand’s proactive tourism policies suggest continued growth in this segment. This article provides an in-depth analysis of the factors influencing Russian tourist arrivals in Thailand and the implications for the future. With a strong foundation of tourism-friendly policies, infrastructure investments, and marketing strategies, Thailand is well-positioned to sustain and expand its appeal to Russian visitors in the coming years.Item Metadata only Jugular venous oximetry(2025-01-01) Lele A.V.; Chaikittisilpa N.; Vavilala M.S.; Lele A.V.; Mahidol UniversityJugular venous oxygen saturation monitoring provides useful information regarding cerebral oxygenation supply and demand. Jugular oximetry can be facilitated by insertion of a catheter by an anesthesiologist or an intensivist. While monitoring neurocritically ill patients to diagnose cerebral hypoxia, jugular venous oximetry can also provide vital information to optimize the relationship between cerebral oxygen demand and supply with the goal to improve patient outcomes.Item Metadata only Co-immunization with recombinant S5196–272 and S6200–317 proteins for enhanced protective antibody response against Tilapia lake virus in Nile tilapia, Oreochromis niloticus (Linnaeus, 1758)(2025-09-01) Plysup B.; Lueangyangyuen A.; Khrisanapant P.; Senapin S.; Rattanarojpong T.; Somsoros W.; Khunrae P.; Sangsuriya P.; Plysup B.; Mahidol UniversityImportance of the work: Co-immunization with recombinant S5196–272 and S6200–317 proteins enhances protective immunity and provides insights for future TiLV vaccine development. Objectives: To evaluate the vaccine potential of combined S5196–272 and S6200–317 compared to individual immunization. Materials and Methods: A sample of Nile tilapia was divided into three main groups: immunized with S5196–272 and S6200–317 individually, or co-immunized. Antibody responses were measured weekly using enzyme-linked immunosorbent assay, with virus neutralization being assessed using a methylthiazolyldiphenyl-tetrazolium bromide (MTT) cell viability assay. A viral challenge test was conducted to determine the relative percentage of survival (RPS). Results: Co-immunization of the fish with S5196–272 and S6200–317 resulted in a synergistic effect, leading to the higher production of S6200–317-specific antibodies than for immunization with S6200–317 alone. A significant increase in serum antibody levels was observed from 7 d, 21 d, 28 d and 35 d post-co-immunization. In contrast, S5196–272-specific antibodies were generated at consistently high levels following both individual and co-immunization. The MTT cell viability assay findings demonstrated that antibodies from the co-immunization group had the highest virus-neutralizing effect (87.22% viability). Furthermore, the viral challenge assay revealed that the co-immunization group had the highest RPS (57.14%), whereas individual immunization provided no protection effect against TiLV infection. Main finding: Co-immunization with S5196–272 and S6200–317 induced a synergistic antibody response and provided effective protection against TiLV in Nile tilapia.Item Metadata only Protective effects of gallic acid against cadmium-induced neuroinflammation in glioblastoma cells.(2025-01-01) Phuagkhaopong S.; Sukwattanasombat J.; Wonganan P.; Vivithanaporn P.; Suknuntha K.; Phuagkhaopong S.; Mahidol UniversityBackground: Environmental exposure to cadmium (Cd), a toxic heavy metal, was associated with an increased incidence risk of glioblastoma multiforme (GBM), the most common, malignant primary brain tumor in adults. One possible mechanism by which Cd exerts its carcinogenic effect is related to inflammation. Our previous studies have demonstrated that Cd induces cytokine response at low concentrations (1-10 µM) and early-time points both in primary human astrocytes and human U-87 MG astrocytes. Gallic acid (GA) is a natural phenolic compound with several therapeutic effects, including anti-inflammatory, antioxidant, and anti-tumor. GA has shown a neuroprotective effect under Cd induction in Wistar rats. However, the underlying mechanism has not been extensively reported. This study aims to investigate the protective effects and potential mechanism of GA against inflammation induced by Cd exposure in human U-87 MG astrocytes. Methods: The effect of GA on Cd-induced expression and release of cytokine was investigated in human astrocytoma U-87 MG astrocytes by real-time PCR and ELISA, respectively as well as molecular mechanisms involved in the anti-inflammatory effect of this drug was investigated by Western Blotting. Results: The results showed that pretreatment of GA suppressed an increased level of interleukin (IL)-6, IL-8, chemokine (C-C motif) ligand (CCL)2, and CCL3 secretion by human U-87 MG astrocytes. In addition, pretreatment of GA inhibited Cd-activated phosphorylation of ERK1/2 MAPK and phosphorylation of signal transducer and activator of transcription 3 (STAT3). Conclusions: These findings suggest that GA may help prevent inflammation-induced GBM.Item Metadata only Enhancing Rice Cultivation Efficiency of Farmers in the Bueng Boraphet Wetland Area, Nakhon Sawan Province, Thailand(2025-10-01) Junhamakasit T.; Atthaboon W.; Anuttarunggoon N.; Taeprayoon P.; Junhamakasit T.; Mahidol UniversityBackground: Agricultural areas surrounding Bueng Boraphet, Nakhon Sawan Province, Thailand, lie outside the irrigation zone, where rice cultivation depends largely on rainfall during the wet season and off-season cropping throughout the year. These conditions complicate water-use assessment and create challenges for sustainable resource management. Major constraints to rice production include water scarcity, high input costs—particularly for fertilizers—limited technical knowledge, improper fertilizer use, frequent pest and disease outbreaks, inefficient weed control, and crop residue burning that degrades soil quality. Collectively, these problems reduce productivity, heighten environmental stress, and undermine long-term sustainability. Objectives and Methodology: This research aimed to enhance rice cultivation efficiency in a 200-rai (32-hectare) pilot area in Wang Mahakon and Thap Krit Sub-districts through participatory action research (PAR). The study integrated site-specific fertilizer management (SSF) with alternate wetting and drying (AWD) water management to optimize both productivity and resource use. The participatory process involved nine key steps: (1) situational analysis and community planning; (2) participatory tools such as “Happiness Compass” and “Smart A4” to identify local needs; (3) establishment of community-led demonstration plots; (4) inter-community learning through study visits; (5) integration of expert knowledge and local wisdom; (6) mutual learning via field visits; (7) participatory feedback and data verification; (8) determination of appropriate field technologies; and (9) soil analysis and fertilizer application based on analytical results. Results and Findings: Results showed that the integrated AWD–SSF system performed significantly better than traditional broadcasting in continuously flooded fields. In transplanted rice plots, the number of tillers per clump averaged 19.94, compared to 4.42 in traditional plots—a nearly fourfold increase—especially within 45 days after planting. This improvement stemmed from enhanced soil aeration during dry intervals, stimulating root and shoot development. Intermittent drying also activated soil microbes, improving nutrient availability and plant vigor. Although both systems showed natural self-thinning around 75 days after planting, overall growth and resilience remained superior in the improved plots. Water consumption in traditional fields averaged 1,351 cubic meters per rai (8,444 m³/ha) per crop cycle, requiring five irrigation events. Under AWD, water use dropped to 810–910 m³/rai (5,060–5,690 m³/ha) with only three irrigation events—representing a 32–40% reduction. Yields increased from 742.32 kg/rai (4.64 t/ha) to 812.33 kg/rai (5.08 t/ha), surpassing the national average and indicating a 9.4% productivity gain. This improvement was attributed to balanced soil fertility management and the avoidance of straw burning, which helped maintain organic matter and nutrient balance. Economic analysis revealed that net returns rose from 2,212.54 THB (≈68 USD) to 2,709.91 THB (≈83 USD) per rai. Although seedling costs were slightly higher, total expenses declined due to lower fertilizer and fuel use coupled with higher yields. The benefit–cost ratio improved from 1.41 to 1.49, strengthening farmers’ incentive to adopt the practice. The transplanting method also improved weed control—water retention during the first month suppressed weed growth by about 70%, and unwanted rice varieties (“weedy rice”) were reduced by 80% through manual removal. The participatory process produced transformative community outcomes. Farmers gained a clearer understanding of production costs, input management, and sustainable practices. Participatory tools like the Happiness Compass encouraged reflection and context-based planning, fostering genuine behavioral change. Demonstration plots became “living laboratories,” where farmers observed biological, economic, and social impacts firsthand. At the socio-economic level, farmers reduced unnecessary expenditures on fertilizers and fuel while improving grain quality and net income. Environmentally, AWD significantly conserved water and reduced chemical runoff, mitigating pollution and supporting the ecological balance of the Bueng Boraphet wetland. Socially, collective learning and leadership were strengthened through the formation of groups such as the Low-Carbon Bueng Boraphet Community Enterprise and Cost-Reduction Learning Groups, which continued collaboration with local agencies and served as community knowledge hubs. The demonstration sites have since evolved into community learning centers, where experienced farmers act as trainers. In recognition of this success, Nakhon Sawan Province issued Provincial Order No. 3387/2567 to establish a steering committee for continued promotion of AWD-based rice farming in the Bueng Boraphet model area. The initiative has inspired inter-subdistrict collaboration through the Water Users and Low-Cost Rice Growers Network, facilitating knowledge sharing and scaling to neighboring areas. The outcomes have also been incorporated into youth training and environmental education curricula, ensuring long-term capacity building and intergenerational learning. Conclusions: Ultimately, this participatory research established a new paradigm for community-based sustainable agriculture. By linking productivity, cost efficiency, environmental stewardship, and quality of life, it strengthened both human and social capital—the essential foundations of sustainability. The Bueng Boraphet experience demonstrates that when local communities actively engage in planning, experimentation, and evaluation, academic innovations such as AWD and SSF can be effectively localized, generating enduring economic, social, and ecological benefits across Thailand’s rainfed rice regions.