Earticle

연구목적: 본 논문에서는 스마트 디바이스의 높은 보급률을 활용하여 근감소증을 추정 및 예측하는 딥 러닝 알고리즘을 제안과 연구를 수행한다. 연구방법: 딥러닝 학습을 위해 스마트 디바이스에 내장된 관 성센서를 활용하여 실험 데이터를 수집하였다. 데이터를 수집하는 테스트용 어플리케이션 구현하여 ‘정상’과 ‘비정상’걸음과 ‘달리기’, ‘낙상’, ‘스쿼트’ 자세의 5 가지 상태를 구분하여 데이터를 수집하였 다. 연구결과: LSTM, CNN, RNN model 사용 시 예측 정확도를 분석했고 CNN-LSTM 융합형 모델을 활 용하여 이진분류 정확도 99.87%, 다중 분류 92.30%의 정확도를 보였다. 결론: 근감소증이 있는 사람의 경우 걸음걸이의 이상이 생긴다는 점에 착안하여 스마트 디바이스를 활용한 연구를 진행하였다. 본 연 구를 활용하여 근감소증으로 인해 생기는 재난안전을 강화 할 수 있을 것이다.

Purpose: In this paper, we propose a study of deep learning algorithms that estimate and predict sarcopenia by exploiting the high penetration rate of smart devices. Method: To utilize deep learning techniques, experimental data were collected by using the inertial sensor embedded in the smart device. We implemented a smart device application for data collection. The data are collected by labeling normal and abnormal gait and five states of running, falling and squat posture. Result: The accuracy was analyzed by comparative analysis of LSTM, CNN, and RNN models, and binary classification accuracy of 99.87% and multiple classification accuracy of 92.30% were obtained using the CNN-LSTM fusion algorithm. Conclusion: A study was conducted using a smart sensoring device, focusing on the fact that gait abnormalities occur for people with sarcopenia. It is expected that this study can contribute to strengthening the safety issues caused by sarcopenia.

Engineering or humanities data are stored in databases and are often used for search services. While the latest deep-learning technologies, such like BART and BERT, are utilized for data analysis, humanities data still rely on traditional databases. Representative analysis methods include n-gram and lexical statistical extraction. However, when using a database, performance limitation is often imposed on the result calculations. This study presents an experimental process using MariaDB on a PC, which is easily accessible in a laboratory, to analyze the impact of the database on data analysis performance. The findings highlight the fact that the database becomes a bottleneck when analyzing large-scale text data, particularly over hundreds of thousands of records. To address this issue, a method was proposed to provide real-time humanities data analysis web services by leveraging the open source database, with a focus on the Seungjeongwon-Ilgy, one of the largest datasets in the humanities fields.

In this study, K-equidistant partitioning (K-EP), a novel data augmentation method, is proposed to address the limitations of sensor data analysis and enhance the performance of behavior classification models. K-EP involves dividing rows of sensor data at equidistant intervals and extracting information from each segment, thereby increasing the size of the dataset by a factor of K. This method is based on the sensor minimum warranted frequency hypothesis, which posits that a sampling frequency of 20-40 Hz provides sufficient data for behavioral classification. The effectiveness of K-EP is validated through three experiments, which involve determining the optimal value of K for K-EP, comparing K-EP with other data augmentation methods, and assessing the added value of K-EP when combined with other methods. The results indicate that K-EP effectively overcomes the quantitative limitations of sensor data and enhances model robustness. It achieves higher F1-scores than existing methods, indicating that it is an effective data augmentation method for sensor-based behavior classification models.

Purpose: This study aimed to identify the main keywords, network structures, and topical themes in patient safety incident reports using text network analysis. Methods: The study analyzed patient safety incident reports from a general hospital in Seoul, covering a total of 3,576 cases reported over five years, from 2019 to 2023. Unstructured data were extracted from the text of the incident reports, detailing how the patient safety incidents occurred and how they were managed according to the six-part principles. The analysis was conducted in four steps: 1) word extraction and refinement, 2) keyword extraction and word network generation, 3) network connectivity and centrality analysis, and 4) topic modeling analysis. The NetMiner program was used for data analysis. Results: The analysis of degree, betweenness, and closeness centrality revealed that the most common keywords among the top five were "confirmation," "medication," "inpatient room," "caregiver," and "condition." Topic modeling analysis identified three main topic groups: 1) incidents caused by a lack of awareness of fall risk, 2) incidents of non-compliance with basic medication principles, and 3) incidents due to inaccurate patient identification. Conclusion: To prevent patient safety incidents, it is necessary to promote a culture of safety in hospitals, standardize patient identification procedures, and provide basic training in medication safety and fall prevention to healthcare staff. Furthermore, empirical research on patient safety practices is necessary to encourage active participation in patient safety activities by patients and family caregivers.

A location-aware algorithm is proposed in this study to optimize the system performance of distributed systems for processing big data with low data reliability and application performance. Compared with previous algorithms, the location-aware data block placement algorithm uses data block placement and node data recovery strategies to improve data application performance and reliability. Simulation and actual cluster tests showed that the location-aware placement algorithm proposed in this study could greatly improve data reliability and shorten the application processing time of I/O interfaces in real-time.

Purpose: Inpatient and residential smoking cessation programs in Korea have demonstrated relatively high quit rates, with residential programs consistently outperforming inpatient ones. However, simple comparisons are limited by differences in participant characteristics and eligibility criteria. This study aimed to determine whether program type independently influences quit rates, using both self-reported and biochemically verified outcomes.Methods: This descriptive study conducted a secondary analysis of data from 17,290 participants enrolled in national smoking cessation services across 18 regional tobacco control centers (2018-2020). Data included demographics, smoking history, and program participation. Quit status at 4 weeks, 12 weeks, and 6 months was assessed through self-report and biochemical verification. Multivariate logistic regression was used to evaluate the independent effect of program type on 6-month quit outcomes.Results: The mean age of participants was 54.8 ± 12.0 years, and 14.5% were women. At 6 months, self-reported quit rates ranged from 16.5% to 34.1% for the inpatient program and from 26.0% to 62.8% for the residential program. Biochemically verified rates ranged from 8.6% to 19.0% (inpatient) and 11.9% to 46.7% (residential). After adjusting for confounders, program type was significantly associated with self-reported quitting (aOR = 0.80; 95% CI = 0.72-0.89; p < .001), but not with biochemically verified quitting (aOR = 0.91; 95% CI = 0.82-1.01; p = .082).Conclusion: Although residential programs showed higher self-reported quit rates, program type did not predict verified cessation. This suggests differences in participant characteristics may drive outcomes. Further research should identify effective, evidence-based components for sustained quitting.

Recently, big data and artificial intelligence (AI) based on communication systems have become one of the hottest issues in the technology sector, and methods of analyzing big data using AI approaches are now considered essential. This paper presents diverse paradigms to subjects which deal with diverse research areas, such as image segmentation, fingerprint matching, human tracking techniques, malware distribution networks, methods of intrusion detection, digital image watermarking, wireless sensor networks, probabilistic neural networks, query processing of encrypted data, the semantic web, decision-making, software engineering, and so on.

Big data information and pattern analysis have applications in many industrial sectors. To reduce energy consumption effectively, the eco-driving method that reduces the fuel consumption of vehicles has recently come under scrutiny. Using big data on commercial vehicles obtained from digital tachographs (DTGs), it is possible not only to aid traffic safety but also improve eco-driving. In this study, we estimate fuel consumption efficiency by processing and analyzing DTG big data for commercial vehicles using parallel processing with the MapReduce mechanism. Compared to the conventional measurement of fuel consumption using the On-Board Diagnostics II (OBD-II) device, in this paper, we use actual DTG data and OBD-II fuel consumption data to identify meaningful relationships to calculate fuel efficiency rates. Based on the driving pattern extracted from DTG data, estimating fuel consumption is possible by analyzing driving patterns obtained only from DTG big data.

Revealing customer satisfaction through big social data has been an interesting research topic in tourism and hospitality. Big data analysis is an effective way to detect customers’ behaviors in their decision-making. This study aims to perform big social data analysis to reveal whether food quality impacts the relationship between hotel performance criteria and travelers’ satisfaction. A two-stage methodology is developed to address the objectives of this study. The findings demonstrated that there is a positive relationship between eco-friendly hotels’ performance criteria and satisfaction. The results and implications for managers and future research directions are discussed.

Freight management systems require a new business model for rapid decision making to improve their business processes by dynamically analyzing the previous experience data. Moreover, the amount of data generated by daily business activities to be analyzed for making better decisions is enormous. Online-to-offline or offline-to-online (O2O) is an electronic commerce (e-commerce) model used to combine the online and physical services. Data analysis is usually performed offline. In the present paper, to extend its benefits to online and to efficiently apply the big data analysis to the freight management system, we suggested a system architecture based on O2O services. We analyzed and extracted the useful knowledge from the real-time freight data for the period 2014-2017 aiming at further business development. The proposed system was deemed useful for truck management companies as it allowed dynamically obtaining the big data analysis results based on O2O services, which were used to optimize logistic freight, improve customer services, predict customer expectation, reduce costs and overhead by improving profit margins, and perform load balancing.

With the rapid growth of genomic data, new requirements have emerged that are difficult to handle with big data storage and analysis techniques. Regardless of the size of an organization performing genomic data analysis, it is becoming increasingly difficult for an institution to build a computing environment for storing and analyzing genomic data. Recently, cloud computing has emerged as a computing environment that meets these new requirements. In this paper, we analyze and compare existing distributed and parallel NGS (Next Generation Sequencing) analysis based on cloud computing environment for future research.

Artificial intelligence, especially deep learning technology, is penetrating the majority of research areas, including the field of bioinformatics. However, deep learning has some limitations, such as the complexity of parameter tuning, architecture design, and so forth. In this study, we analyze these issues and challenges in regards to its applications in bioinformatics, particularly genomic analysis and medical image analytics, and give the corresponding approaches and solutions. Although these solutions are mostly rule of thumb, they can effectively handle the issues connected to training learning machines. As such, we explore the tendency of deep learning technology by examining several directions, such as automation, scalability, individuality, mobility, integration, and intelligence warehousing.

This paper investigates Korean drama viewing motivation's impact on Chinese consumers' cosmetics purchase intention using PLS-SEM with 458 valid responses. Five motivation dimensions (Cultural Curiosity, Drama Appeal, Cultural Proximity, Interpersonal Influence, Viewing Habits) were examined through two mediating variables: Korean Cultural Affinity and Korean Visitation Intention. All twelve hypotheses were significant. Cultural Curiosity demonstrated the strongest effect on Korean Cultural Affinity (β = 0.318), while Interpersonal Influence most impacted Korean Visitation Intention (β = 0.400). Notably, Korean Cultural Affinity exhibited greater influence on purchase intention (β = 0.501) than Korean Visitation Intention (β = 0.256), revealing that cultural connection supersedes travel interest in driving cosmetics consumption. The model achieved 49.5% explanatory power for purchase intention variance. This research fills a gap by identifying specific consumption mechanisms beyond general Korean Wave influence offering strategic insights for Korean cosmetics brands targeting Chinese markets.

Along with the rapid development of the economy, the urban scale has extended rapidly, leading to the formation of different types of urban function districts (UFDs), such as central business, residential and industrial districts. Recognizing the spatial distributions of these districts is of great significance to manage the evolving role of urban planning and further help in developing reliable urban planning programs. In this paper, we propose an automatic UFD division method based on big data analysis of point of interest (POI) data. Considering that the distribution of POI data is unbalanced in a geographic space, a dichotomy-based data retrieval method was used to improve the efficiency of the data crawling process. Further, a POI spatial feature analysis method based on the mean shift algorithm is proposed, where data points with similar attributive characteristics are clustered to form the function districts. The proposed method was thoroughly tested in an actual urban case scenario and the results show its superior performance. Further, the suitability of fit to practical situations reaches 88.4%, demonstrating a reasonable UFD division result.

Purpose: Although digital solutions could mitigate the challenges faced by older adults living alone(OALA), only a few studies investigated the need for and acceptance of digital health interventions forself-management (DHISMs) among this demographic. Thus, we aim to explore this need and acceptance,along with the contextual factors, among OALA. Methods: A mixed-methods research approach was adopted. We conducted 1) a quantitative survey(n ¼ 191) to investigate the need for and acceptance of DHISMs using a numeric rating scale and 2) aqualitative study (n ¼ 24) based on focus group interviews to explore contextual factors related to thequantitative results. Qualitative data were analyzed using thematic analysis. Results: In the quantitative study, the mean scores for the need for and acceptance of DHISMs were 6.41and 6.53 out of 10, respectively. Emergency response systems had the highest need and acceptancescores, whereas digital interventions for behavioral change (medication adherence, sleep, stress, and dietmanagement) had relatively lower scores. The qualitative analysis revealed two themes and five subthemes:the need for inclusive support for independent living (environmental safety and diverseself-management challenges with limited support) and multidimensional factors related to DHISMacceptance (personal, technological, and relational barriers and facilitators). Conclusions: In the future, the unique and multidimensional factors influencing the need for andacceptance of DHISMs among OALA should be carefully considered to support their self-managementand independent living. Blended care, which involves integrating age-friendly technology withpersonalized human interaction, is pivotal for increasing DHISM acceptance in this population.

Purpose: This study investigated the incidence of urinary tract infection (UTI) in community-dwelling adults and identified the association between obesity and UTI. Methods: The participants were 4,926 adults aged over 40 years who had no UTIs at the baseline survey of the Korean Genome Epidemiology Study. Obesity was defined according to the cirtieria of Korean Society for the Study of Obesity using body mass index (BMI) data. UTI was defined as those who had self-reported UTI or had either nitrite, or both leukocytes and blood in the urine dipstick test. Hazard ratio (HR) and 95% confidence interval (CI) were calculated using a multivariate Cox proportional hazards regression analysis to identify the association between the obesity and UTI. Results: The incidence proportion of UTI was 5.1%, and the incidence density per 1,000 person-years was 25.5. After controlling general characteristics, people with BMI ≥30.0 kg/m² remained 1.66 times (HR = 1.66, 95% CI = 1.06~2.60; p < .05) more likely to have UTI than those with normal weight. This trend was also present in men or people aged ≥ 60 years. Among women aged ≥ 60 years, people with BMI ≥ 30.0 kg/m² were 1.98 times (HR = 1.98, 95% CI = 1.01~3.86; p < .05) more likely to have UTI than those with normal weight. Conclusion: The BMI ≥ 30.0 kg/m² is a risk factor of UTIs in Korean adult men over 40 years and women aged ≥ 60 years. It is necessary to emphasize the importance of obesity management to men or women aged ≥ 60 years, specifically.

Purpose: Self-rated health in individuals with chronic diseases is influenced by various factors, including dietary adherence and physical activity. However, limited research has investigated how these factors relate to self-rated health among people with chronic kidney disease. Therefore, this study aimed to describe self-rated health and identify its associated factors in this population. Methods: This cross-sectional, secondary data analysis utilized datasets from the seventh Korea National Health and Nutrition Examination Survey, which were collected between 2016 and 2018. A total of 557 participants (mean age=74.8 years) with a glomerular filtration rate of <60 mL/min/1.73 m² were included. Data from health interviews and examinations were analyzed to assess self-rated health, dietary adherence, and physical activity. Descriptive and inferential statistical methods were employed for analysis. Results: Among the 557 participants, 42.6% rated their health as poor. Factors such as sex, age, income, smoking history, anxiety/depression, number of comorbidities, glomerular filtration rate, and physical activity were significantly associated with self-rated health. In contrast, dietary adherence did not exhibit a significant association. Conclusion: Understanding the factors associated with self-rated health can inform the development of nursing interventions aimed at improving self-rated health among patients with chronic kidney disease.

Purpose: This study aimed to investigate the association among various adolescent prob- lem behaviors in South Korea, focusing on gender differences. Methods: This secondary analysis was conducted using cross-sectional data from the 19th Korea Youth Risk Behavior Survey, conducted in 2023 which included 52,880 mid- dle and high school students in South Korea. They completed an anonymous self-admin- istered survey regarding problem behaviors (drinking alcohol, smoking, drugs use, and sexual experiences). Data were analyzed using complex-samples chi-square and multiple logistic regression models. Results: Among the 52,880 adolescents, the prevalence rates of alcohol use, smoking, drug use, and sexual experiences were 32.6%, 8.6%, 1.7%, and 6.5%, respectively. Gender differences were observed in alcohol use complex-samples multiple logistic regression models. However, no significant gender difference was found in drug use (p=.250). Four problem behaviors were significantly associated with one other when analyzed as inde- pendent variables (odds ratio [OR], 1.33?10.85). The strongest associations were found between alcohol use and smoking (OR, 10.49?10.85), and between smoking and sexual experiences (OR, 4.91?4.96). Conclusion: This study found significant gender differences in adolescent problem be- haviors, with male adolescents exhibiting higher rates of alcohol use, smoking, and sexu- al experience. Strong associations were observed between alcohol use and smoking, as well as between smoking and sexual experience. These findings suggest the need for in- tegrated intervention strategies that target multiple co-occurring problem behaviors.

Purpose: The majority of adolescents use smartphones, and their overdependence on smartphones has emerged as a serious social issue. Although studies have explored the effect of smartphone overdependence on adolescent problem behaviors, research on its influence on sexual behavior is scarce. This study aimed to examine the association be- tween smartphone overdependence and sexual behavior among adolescents. Methods: This study used data from the 19th Korea Youth Risk Behavior Web-Based Survey conducted in 2023. Smartphone overdependence was the independent variable, and sexual behaviors (sexual intercourse and contraceptive use) were the dependent variables. Multivariate regression analysis was performed to control for confounding variables. Results: The study participants included 52,880 adolescents aged 13?18 years. Among them, 28.0% (n=14,672) were classified as smartphone overdependent, 6.5% (n=3,349) had sexual experience, and 65% (n=2,182) of those with sexual experience reported using contraception. Smartphone overdependence was not significantly associated with sexual experience but was significantly associated with contraceptive use. Adolescents in the non-overdependent group were 1.27 times more likely to use contraception than those in the smartphone overdependent group (95% confidence interval, 1.07?1.52; p<.007). Conclusion: The findings highlight a significant association between smartphone over- dependence and contraceptive use among adolescents. Policy interventions and educa- tional strategies that consider adolescents’ smartphone usage patterns and trait factors are necessary.

Purpose: The purpose of this study is to identify the factors influencing generalized anxiety disorder (GAD) in older adults and to provide effective health management directions for GAD patients. Methods: This study used data from the 9th National Health and Nutrition Survey (2022) and selected 1,666 older adults. The analysis was conducted using the SPSS 26.0 program. A complex sample chi-square test was performed to identify factors related to GAD. Complex sample logistic regression was conducted to determine the factors influencing GAD in the participants. Results: The study results indicated that sex (p=.034), smoking (p=.042), alcohol consumption (p=.003), sleep duration (p=.005), chewing discomfort (p<.001), speaking discomfort (p<.001), and self-rated health (p=.016) were statistically significantly associated with GAD. The multivariate analysis considering the interaction between variables, sleep duration (OR=2.07, p=.031) and chewing discomfort (OR=2.49, p=.008) were identified as factors influencing GAD. Conclusion: Chewing discomfort in older adults affects not only GAD but also leads to a decline in aesthetics and social withdrawal. Therefore, improvements in oral health policy support for oral health promotion and prevention are required. Additionally, education and interventions through community systems should be provided to address sleep issues associated with GAD.