Buku & Publikasi Ilmiah

This research employs a mixed-methods design to analyze telemedicine adoption in rural Indonesia by integrating the Technology Acceptance Model (TAM) and the Information System Success Model (ISS), extended with digital literacy and usability. A quantitative survey of 314 respondents was complemented by in-depth interviews with 50 participants and demographic analysis using chi-square and logistic regression. The quantitative findings reveal that the primary adoption construct is Usability → Perceived Ease of Use (PEOU) → Perceived Usefulness (PU) → Intention to Use (ITU) → Net Benefit (NB). Perceived usefulness emerged as the strongest predictor of both satisfaction and intention. Information Quality significantly influenced satisfaction, whereas System Quality did not, indicating that clear medical information outweighs technical system performance in shaping satisfaction. Similarly, usability directly did not affect PU but indirectly through PEOU, and digital literacy influenced PU but not PEOU. Demographic analysis confirmed that occupation was significant—students and healthcare workers acted as early adopters—while age and prior training were not, suggesting that adoption transcends generational boundaries due to the widespread penetration of the JKN Mobile platform. Qualitative insights enriched these findings by highlighting key barriers and enablers such as inconsistent interfaces, infrastructure limitations, privacy concerns, community-based socialization, and expectations for adaptive features like AI diagnostics and pharmacy integration. Overall, the research confirms that telemedicine adoption in rural Indonesia is shaped by the synergy of usability, digital literacy, information quality, and social context rather than by training or demographic variables alone.

The rapid growth of online gaming has raised concerns about addictive behaviors among young people, particularly with the emergence of loot boxes that resemble gambling mechanisms. This study aims to examine the effectiveness of a gamification-based application as a preventive intervention for online gambling game addiction and to evaluate user acceptance through an extended Technology Acceptance Model (TAM). The research was conducted in two stages. In the pre-test phase, 588 respondents aged 15–25 completed a questionnaire measuring impulsivity, Internet Gaming Disorder (IGD), and loot box exposure. The results identified 169 individuals (28.7%) with addictive tendencies. In the intervention phase, 86 respondents from this group participated in a gamified stimulation program using a specially designed application. Data were analyzed using Partial Least Squares Structural Equation Modeling (PLS-SEM). The measurement model met reliability and validity requirements. Structural model analysis confirmed the classic TAM relationships: perceived ease of use significantly influenced perceived usefulness (β = .559, p < 0.001), perceived usefulness influenced attitude toward use (β = .385, P= 0.001), and attitude influenced behavioral
intention (β = .461, p < 0.001). In addition, Self-Determination Theory (SDT) significantly affected both attitude (β = .360, P= 0.003) and behavioral intention (β = .166, P= 0.038). However, Affective Visual Design (AVD) was not significant, and behavioral intention did not reduce addictive behavior (β = -0.109, P= 0.386). The model demonstrated predictive relevance for PU, ATT, and BI (Q² > 0), whereas ABT showed no predictive relevance (Q² = –0.003), indicating limited short-term behavioral predictability. This study contributes theoretically by extending TAM with SDT in the context of digital health interventions and practically by demonstrating the potential of gamification as a preventive tool. However, the short intervention period and limited sample size constrained its effectiveness in reducing addiction. Longer-term interventions and broader contextual factors are recommended for future research.

The limited human resources in dental clinics often lead to delays in responding to patients' initial consultations, potentially reducing service quality and patient satisfaction. The advancement of Artificial Intelligence (AI) has encouraged the adoption of chatbots based on Natural Language Processing (NLP) to improve information service efficiency. This study aims to implement a chatbot using the Bidirectional Encoder Representations from Transformers (BERT) model to enhance consultation efficiency at the dental clinic of drg. Fahmi Nurdin. The system development method used is Agile, with stages including requirement analysis, user interface design, model training, and chatbot implementation. The dataset consists of 49 tags (topics) and 490 patterns (questions), formatted in JSON, and used to train the BERT model for intent classification and response generation. An accuracy score of 0.86 was obtained, along with a macro precision of 0.867, macro recall of 0.857, and a macro F1-score of 0.839, indicating that the model is capable of performing classification with a relatively balanced performance across all classes. Evaluation results show that the chatbot can provide relevant and accurate responses, accelerating the initial consultation process. This implementation is expected to help overcome human resource limitations, improve service quality, and strengthen patient trust in the clinic's services.

The attendance system plays a crucial role in supporting the learning process and academic administration in various institutions, including the Faculty of Computer Science, Kuningan University. Currently, the attendance system used still relies on conventional methods, namely hard copy-based attendance sheets signed by students. This method has several disadvantages, such as a long time in recapitulating data to the academic system, the potential for signature manipulation, and the risk of losing or damaging documents. To overcome
these problems, this research proposes the application of face recognition technology based on Convolutional Neural Network (CNN). CNN is a development of Multi-layer Perceptron (MLP) specifically designed to process two-dimensional data, such as facial images. This research uses a prototype development model consisting of three main stages: (1) user needs analysis, (2) mock-up design, and (3) design testing. The face identification process produces a feature description that represents the characteristics of the detected face. This feature description is then compared with the face data stored in the database to perform matching. The test results using the K-Fold method show that the proposed model achieves an accuracy of 92%. Thus, this face recognition-based attendance system is expected to increase efficiency, reduce the potential for fraud, and provide a more modern and integrated attendance solution for the Faculty of Computer Science, Kuningan University.

Inter Sport Store, a retail business located in Kersana District, Brebes Regency, has been operating since 2017. The store faces ongoing challenges in inventory management, frequently encountering stock surpluses or shortages that result in losses of up to IDR 15 million per month. These issues primarily result from a manual ordering process that relies on subjective estimation, without the support of accurate data analysis. Furthermore, sales records are still maintained manually in notebooks, making it difficult to track data and analyze consumer demand effectively. This study proposes the development of a web-based sales forecasting system using the Triple Exponential Smoothing (TES) method, which can handle both trend and seasonal patterns simultaneously, providing higher accuracy for complex time-series data compared to Single or Double Exponential Smoothing methods. The system is developed using the prototype development method with UML design and implemented using Laravel framework and MySQL database for real-time access and data updates. Forecasting results for the Ortuseight Catalyst Legion FG – Aqua White soccer shoe yielded a Mean Absolute Percentage Error (MAPE) of 13.38%, which is considered within the "good" accuracy range. The implementation of this system successfully reduced overstock by 25% and improved inventory management efficiency. This forecasting system enhances operational efficiency and supports data-driven decision-making for store management.

The Mekar Jaya Savings and Loan Cooperative (Luragung Branch) faces issues with non-performing loans stemming from creditworthiness assessments that remain subjective, relying on the "Five C’s of Credit" method. This study aims to develop a website for predicting customer credit risk by implementing the Naïve Bayes algorithm, enabling the cooperative to assess risk more objectively and accurately. The website was built using the Prototyping system development method, utilizing JavaScript, the Next.js framework, and a MySQL database. The dataset comprises 500 customer credit records, including attributes such as marital status, income, loan type, loan amount, installment amount, loan term, and collateral, alongside risk labels (high-risk vs. low-risk). The data was split into a training set of 350 records and a testing set of 150 records. Testing using a confusion matrix demonstrated that the Naïve Bayes algorithm achieved an accuracy rate of 98%.

Dengue fever is one of the most common infectious diseases in Indonesia. However, information regarding its prevention and treatment remains fragmented, with limited accessibility and availability. This study developed a chatbot integrating the Long ShortTerm Memory (LSTM) algorithm to answer various questions about dengue fever, including its symptoms, prevention, and treatment. The dataset used consists of questions and answers related to dengue fever, sourced from both primary and secondary data. The data undergoes a series of preprocessing steps before being used for model development, training, and evaluation. The test results indicate that the developed model achieved an accuracy of 100% during validation with a loss function value of 0.0221. These findings demonstrate that the LSTM-based chatbot can provide accurate and relevant responses, making it an effective tool for educating the public in an interactive and efficient manner. This implementation is expected to offer an innovative solution for increasing public awareness of dengue fever prevention and management

Technological developments, including the internet, and learning opportunities are increasing. This also encourages the development of learning strategies and models. The blended learning model is applied in almost all universities in Indonesia and the world. With so many universities in Indonesia, implementing blended learning is a challenging thing because it requires a lot of technological preparation and human resources. This research aims to identify factors, develop a model, and evaluate the model to see the readiness and acceptance of technology for adopting social media in blended learning among private higher education institutes students in Indonesia. The population of this research is students from private higher education institutes in West Java, Indonesia, who are directly involved in using blended learning and social media. This quantitative research used a research instrument with five-Likert’s scale. The research population was 663,307, with a sample of 384 students spread across West Java. The contribution of this research is to make a significant contribution to the theoretical framework by expanding and refining existing concepts, providing a more comprehensive understanding of the readiness and acceptance factors for the adoption of social media in blended learning so that it has the potential to provide information to learning planners at private higher education institutes in West Java, Indonesia to help make the right decisions and optimize blended learning planning using social media technology. These findings statistically explain that 19 of 31 the hypotheses are the accepted ones. Moreover, nine of 12 variables influenced the readiness and acceptance of social media technology in blended learning based on the student perception among the private higher education institutions. They were the technological literacy factor, perceived validity, perceived trust, and technology readiness factors, namely optimism and Innovativeness, and technology acceptance factors, namely perceived effectiveness, perceived easy to use, intention to use and usage behaviour.

Stunting among children under five remains a critical global health concern, particularly in rural areas with limited access to comprehensive health services. The adoption of Electronic Medical Records (EMR), exemplified by the e-PPGBM application used at Public Health Centers (PHCs) in Indonesia, holds substantial promise for improving the monitoring of under-five nutritional status and early identification of stunting risks. Despite its potential, the effectiveness of EMR systems like e-PPGBM in rural PHCs has not been widely assessed. This study evaluates the performance of the e-PPGBM application using the ISO 25010:2011 software quality standard, focusing on six key characteristics: functionality, performance efficiency, usability, security, reliability, and maintainability. A multi-method approach involving black-box testing, scenario testing, stress testing, and usability questionnaires was applied, with responses gathered from 37 PHC personnel. Findings indicate that, while infrastructure and training limitations persist, the e-PPGBM application demonstrates overall effectiveness in tracking nutritional status and detecting stunting risk in Kuningan District, West Java, Indonesia. Notably, the application contributes to improved health service delivery. This study suggests the integration of predictive analytics for stunting risk, based on additional data from local health posts (posyandu), to enhance future usability and performance.

Artificial Intelligence (AI) is increasingly transforming educational practices by enabling more adaptive and personalized learning experiences in secondary schools. Nevertheless, previous applications of the Human–Organization–Technology Fit (HOT-Fit) model have given limited attention to the roles of information literacy and trust in influencing AI adoption. To address this gap, the present study expands the HOT-Fit framework by incorporating three additional constructs: information literacy, perceived validity, and perceived trust, in order to better explain AI readiness and adoption in educational settings. A quantitative approach was employed involving 316 senior high school students from Kuningan Regency, Indonesia. Data
were gathered using a structured questionnaire based on a five-point Likert scale and analyzed through Partial Least Squares Structural Equation Modeling (PLS-SEM) with SmartPLS v3.0 to evaluate 29 proposed hypotheses. The findings revealed that 21 hypotheses were statistically supported. Information literacy demonstrated a strong positive effect on perceived trust (β = 0.708; p < 0.001), as well as on system use, organizational structure, environmental support, and user satisfaction. In addition, system quality significantly contributed to user satisfaction, whereas service quality affected both system use and satisfaction. Among all relationships, net benefit exerted the strongest effect on action to use (β = 0.547; p < 0.001). The R² results for several endogenous constructs were above 0.50, indicating acceptable explanatory capability of the proposed model. Practically, the findings offer implications for educators, policymakers, and system developers in designing AI-supported learning environments by emphasizing the enhancement of digital literacy, service support, and system effectiveness for sustainable AI integration in schools.