• Korean Journal of Radiology
• /
• v.22 no.7
• /
• pp.1213-1224
• /
• 2021

Objective: To develop a machine learning (ML) pipeline based on radiomics to predict Coronavirus Disease 2019 (COVID-19) severity and the future deterioration to critical illness using CT and clinical variables. Materials and Methods: Clinical data were collected from 981 patients from a multi-institutional international cohort with real-time polymerase chain reaction-confirmed COVID-19. Radiomics features were extracted from chest CT of the patients. The data of the cohort were randomly divided into training, validation, and test sets using a 7:1:2 ratio. A ML pipeline consisting of a model to predict severity and time-to-event model to predict progression to critical illness were trained on radiomics features and clinical variables. The receiver operating characteristic area under the curve (ROC-AUC), concordance index (C-index), and time-dependent ROC-AUC were calculated to determine model performance, which was compared with consensus CT severity scores obtained by visual interpretation by radiologists. Results: Among 981 patients with confirmed COVID-19, 274 patients developed critical illness. Radiomics features and clinical variables resulted in the best performance for the prediction of disease severity with a highest test ROC-AUC of 0.76 compared with 0.70 (0.76 vs. 0.70, p = 0.023) for visual CT severity score and clinical variables. The progression prediction model achieved a test C-index of 0.868 when it was based on the combination of CT radiomics and clinical variables compared with 0.767 when based on CT radiomics features alone (p < 0.001), 0.847 when based on clinical variables alone (p = 0.110), and 0.860 when based on the combination of visual CT severity scores and clinical variables (p = 0.549). Furthermore, the model based on the combination of CT radiomics and clinical variables achieved time-dependent ROC-AUCs of 0.897, 0.933, and 0.927 for the prediction of progression risks at 3, 5 and 7 days, respectively. Conclusion: CT radiomics features combined with clinical variables were predictive of COVID-19 severity and progression to critical illness with fairly high accuracy.

• Korea journal of population studies
• /
• v.11 no.1
• /
• pp.97-116
• /
• 1988

This study examines the socioeconomic determinants of fertility behavior in Korea by developing a model which simultaneously takes into account both individual and community-level differences. It especially focuses on the micro-macro nexus of fertility behavior depending on social contexts. This study utilizes micro data obtained from the 1974 Korean National Fertility Survey(KNFS), and macro data obtained from Korean government statistics. The framework of the model is formalized as a set of structural equations modelling the fertility process. The model is formed on a cohort-specific processual basis and is restricted to five-year birth cohorts. Three cohorts of women are studied : those aged 30-34, 35-39, and 40-44. The model includes three fertility-process components : age at first birth, early fertility, and later fertility, which are defined by reference to the age of the mother. The results of this study indicate that socioeconomic development in Korea results in increased age at first birth and reduced numbers of children per couple. In addition to the developmental change, Korea's fertility decline is found to be facilitated by family planning programs. As expected, the effect of family planning on fertility is greater among better-educated women than among poorly educated women. The inconsistent but suggestive result, however, is that the effect of socioeconomic development on fertility is greater among less-privileged women than among more-previleged women.

• International Journal of Advanced Culture Technology
• /
• v.7 no.1
• /
• pp.8-13
• /
• 2019

Despite technical advances in healthcare, the rates of hospital-acquired pressure injury (HAPI) are still high although many are potentially preventable. The purpose of this study was to determine whether tree-based prediction modeling is suitable for assessing the risk of HAPI in ICU patients. Retrospective cohort study has been carried out. A decision tree model was constructed with Age, Weight, eTube, diabetes, Braden score, Isolation, and Number of comorbid conditions as decision nodes. We used RStudio for model training and testing. Correct prediction rate of the final prediction model was 92.4 and the Area Under the ROC curve (AUC) was 0.699, which means there is about 70% chance that the model is able to distinguish between HAPI and non-HAPI. The results of this study has limited generalizability as the data were from a single academic institution. Our research finding shows that the data-driven tree-based prediction modeling may potentially support ICU sensitive risk assessment for HAPI prevention.

• Genomics & Informatics
• /
• v.20 no.2
• /
• pp.16.1-16.12
• /
• 2022

Various methodologies for the genetic analysis of longitudinal data have been proposed and applied to data from large-scale genome-wide association studies (GWAS) to identify single nucleotide polymorphisms (SNPs) associated with traits of interest and to detect SNP-time interactions. We recently proposed a grid-based Bayesian mixed model for longitudinal genetic data and showed that our Bayesian method increased the statistical power compared to the corresponding univariate method and well detected SNP-time interactions. In this paper, we further analyze longitudinal obesity-related traits such as body mass index, hip circumference, waist circumference, and waist-hip ratio from Korea Association Resource data to evaluate the proposed Bayesian method. We first conducted GWAS analyses of cross-sectional traits and combined the results of GWAS analyses through a meta-analysis based on a trajectory model and a random-effects model. We then applied our Bayesian method to a subset of SNPs selected by meta-analysis to further discover SNPs associated with traits of interest and SNP-time interactions. The proposed Bayesian method identified several novel SNPs associated with longitudinal obesity-related traits, and almost 25% of the identified SNPs had significant p-values for SNP-time interactions.

• Health Policy and Management
• /
• v.27 no.2
• /
• pp.149-156
• /
• 2017

Background: This study was conducted to evaluate the performance of the Hierarchical Condition Category (HCC) model, identify potentially high-cost patients, and examine the effects of adding prior utilization to the risk model using Korean claims data. Methods: We incorporated 2 years of data from the National Health Insurance Services-National Sample Cohort. Five risk models were used to predict health expenditures: model 1 (age/sex groups), model 2 (the Center for Medicare and Medicaid Services-HCC with age/sex groups), model 3 (selected 54 HCCs with age/sex groups), model 4 (bed-days of care plus model 3), and model 5 (medication-days plus model 3). We evaluated model performance using $R^2$ at individual level, predictive positive value (PPV) of the top 5% of high-cost patients, and predictive ratio (PR) within subgroups. Results: The suitability of the model, including prior use, bed-days, and medication-days, was better than other models. $R^2$ values were 8%, 39%, 37%, 43%, and 57% with model 1, 2, 3, 4, and 5, respectively. After being removed the extreme values, the corresponding $R^2$ values were slightly improved in all models. PPVs were 16.4%, 25.2%, 25.1%, 33.8%, and 53.8%. Total expenditure was underpredicted for the highest expenditure group and overpredicted for the four other groups. PR had a tendency to decrease from younger group to older group in both female and male. Conclusion: The risk adjustment models are important in plan payment, reimbursement, profiling, and research. Combined prior use and diagnostic data are more powerful to predict health costs and to identify high-cost patients.

• Journal of Korean Biological Nursing Science
• /
• v.19 no.3
• /
• pp.123-130
• /
• 2017

Purpose: This study aimed to identify factors influencing intra-operative core body temperature (CBT), and to develop a predictive model for intra-operative CBT in laparoscopic abdominal surgery. Methods: The prospective observational study involved 161 subjects, whose age, weight, and height were collected. The basal pre-operative CBT, pre-operative blood pressure, and heartbeat were measured. CBT was measured 1 hour and 2 hours after pneumoperitoneum. Results: Explanatory factors of intra-operative hypothermia (< $36^{\circ}C$) were weight (${\beta}=.361$, p< .001) and pre-operative CBT (${\beta}=.280$, p= .001) 1 hour after pneumoperitoneum (Adjusted $R^2=.198$, F= 7.56, p< .001). Weight was (${\beta}=.423$, p< .001) and pre-operative CBT was (${\beta}=.206$, p= .011) 2 hours after pneumoperitoneum (Adjusted $R^2=.177$, F= 5.93, p< .001). The researchers developed a predictive model for intra-operative CBT ($^{\circ}C$) by observing intra-operative CBT, body weight, and pre-operative CBT. The predictive model revealed that intra-operative CBT was positively correlated with body weight and pre-operative CBT. Conclusion: Influence of weight on intra-operative hypothermia increased over time from 1 hour to 2 hours after pneumoperitoneum, whereas influence of pre-operative CBT on intraoperative hypothermia decreased over time from 1 hour to 2 hours after pneumoperitoneum. The research recommends pre-warming for laparoscopic surgical patients to guard against intra-operative hypothermia.

• Asian Pacific Journal of Cancer Prevention
• /
• v.14 no.3
• /
• pp.2013-2017
• /
• 2013

Background: To compare the effects of two adjuvant chemotherapy regimens, anthracycline-based and cyclophosphamide, methotrexate, fluorourical (CMF) on disease free survival for breast cancer patients in the Eastern Mediterranean region and Asia. Methods: In a systematic review with a multivariate mixed model meta-analysis, the reported survival proportion at multiple time points in different studies were combined. Our data sources were studies linking the two chemotherapy regimens on an adjuvant basis with disease free survival published in English and Persian in the Eastern Mediterranean region and Asia. All survival curves were generated with Graphdigitizer software. Results: 14 retrospective cohort studies were located from electronic databases. We analyzed data for 1,086 patients who received anthracycline-based treatment and 1,109 given CMF treatment. For determination of survival proportions and time we usesb the transformation Ln (-Ln(S)) and Ln (time) to make precise estimations and then fit the model. All analyses were carried out with STATA software. Conclusions: Our findings showed a significant efficacy of anthracycline-based adjuvant therapy regarding disease free survival of breast cancer. As a limitation in this meta-analysis we used studies with different types of anthracycline-based regimens.

• Proceedings of the Korea Contents Association Conference
• /
• 2016.05a
• /
• pp.383-384
• /
• 2016

The Special Needs Education Assessment Tool (SNEAT) were verified of reliability and validity. However, the reliability and validity has been verified is only Okinawa Prefecture, the national data has not been analyzed. Therefore, this study aimed to verify the reliability and construct validity of SNEAT in Miyagi Prefecture as part of the national survey. SNEAT using 55 children collected from the classes on independent activities of daily living for children with disabilities in Miyagi Prefecture between November and December 2015. Survey data were collected in a longitudinal prospective cohort study. The reliability of SNEAT was verified via the internal consistency method; the coefficient of Cronbach's ${\alpha}$ were over 0.7. The validity of SNEAT was also verified via the latent growth curve model. SNEAT is valid based on its goodness-of-fit values obtained using the latent growth curve model, where the values of comparative fit index (0.997), tucker-lewis index (0.996) and root mean square error of approximation (0.025) were within the goodness-of-fit range. These results indicate that SNEAT has high reliability and construct validity.

• Communications for Statistical Applications and Methods
• /
• v.29 no.3
• /
• pp.287-299
• /
• 2022

In radiation epidemiology, the excess relative risk (ERR) model is used to determine the dose-response relationship. In general, the dose-response relationship for the ERR model is assumed to be linear, linear-quadratic, linear-threshold, quadratic, and so on. However, since none of these functions dominate other functions for expressing the dose-response relationship, a Bayesian semiparametric method using splines has recently been proposed. Thus, we improve the Bayesian semiparametric method for the selection of the tuning parameters for splines as the number and location of knots using a Bayesian knot selection method. Equally spaced knots cannot capture the characteristic of radiation exposed dose distribution which is highly skewed in general. Therefore, we propose a nonparametric Bayesian knot selection method based on a Dirichlet process mixture model. Inference of the spline coefficients after obtaining the number and location of knots is performed in the Bayesian framework. We apply this approach to the life span study cohort data from the radiation effects research foundation in Japan, and the results illustrate that the proposed method provides competitive curve estimates for the dose-response curve and relatively stable credible intervals for the curve.

• Journal of Preventive Medicine and Public Health
• /
• v.32 no.4
• /
• pp.452-458
• /
• 1999

Objectives: The authors conducted the study to evaluate bias when potentially diseased subjects were included in cohort members while analyzing risk factors of chronic liver diseases. Methods: Total of 14,529 subjects were followed up for the incidence of liver diseases from January 1993 to June 1997. We have used databases of insurance company with medical records, cancer registry, and death certificate data to identify 102 incident cases. The cohort members were classified into potentially diseased group(n=2,217) when they were HBsAg positive, serum GPT levels higher than 40 units, or had or has liver diseases in baseline surveys. Cox's model were used for potentially diseased group, other members, and total subjects, respectively. Results: The risk factors profiles were similar for total and potentially diseased subjects: HBsAg positivity, history of acute liver disease, and recent quittance of smoking or drinking increased the risk. while intake of pork and coffee decreased it. For the potentially diseased, obesity showed marginally significant protective effect. Analysis of subjects excluding the potentially diseased showed distinct profiles: obesity increased the risk, while quitting smoking or drinking had no association. For these intake of raw liver or processed fish or soybean paste stew increased risk; HBsAg positivity, higher levels of liver enzymes and history of acute liver diseases increased the risk. Conclusions: The results suggested the potential bias in risk ratio estimates when potentially diseased subjects were included in cohort study on chronic liver diseases, especially for lifestyles possibly modified after disease onset. The analytic strategy excluding potentially diseased subjects was considered appropriate for identifying risk factors for chronic liver diseases.