Kyeong Eun Lee

Bayesian variable selection under the proportional hazards mixed-effects model

Over the past decade much statistical research has been carried out to develop models for correlated survival data; however, methods for model selection are still very limited. A stochastic search variable selection (SSVS) approach under the proportional hazards mixed-effects model (PHMM) is developed. The SSVS method has previously been applied to linear and generalized linear mixed models, and to the proportional hazards model with high dimensional data. Because the method has mainly been developed for hierarchical normal mixture distributions, it operates on the linear predictor under the Cox type models. The PHMM naturally incorporates the normal distribution via the random effects, which enables SSVS to efficiently search through the candidate variable space. The approach was evaluated through simulation, and applied to a multi-center lung cancer clinical trial data set, for which the variable selection problem was previously debated upon in the literature.

Bootstrap method to evaluate tightness of clusters with application to the Korean standard occlusion study

Dental arch form is an important part of dental orthodontic practice. Distance-based clustering methods are often used to find standard arch forms. In particular, S-J. Lee, S.I. Lee, J. Lim, H-J. Park, and T. Wheeler Method to classify human dental arch form, Am. J. Orthod. Dentofacial Orthop. (2010), to appear] propose a 1-type distance which is invariant to the location-shift and the rotational transformation. Despite the popularity of the distance-based methods, little attention is given to the choice of the distance which has a great influence on final clusters. We have three goals in this paper. First, we study the properties of the 1-type distance by Lee et al. (2010). Second, we propose a bootstrap-based procedure to evaluate quantitatively how good the clusters are. Finally, we apply the bootstrap procedure to the Korean standard occlusion study and compare the existing distance-based clustering methods in previous literature.

A Regression Model for the AUC of Clustered Ordinal Test Results and Working Independent Optimal Weights

We study a regression model on the area under the receiver operating characteristic curves (AUC) for clustered (or repeatedly measured) test results. To account for cluster information, we consider a weighted estimating equation for Dodd and Pepe (2003)s regression model with working independence weights. We find the optimal weight in the given class of working independence weights to minimize the variance (or MSE) of regression estimators. We apply the proposed procedure to analyzing our recent experiment on diagnosing a liver disorder. In this experiment, we investigated MRI images of patients having symptoms of potential liver disorder to compare the performance of different MRI picturing methods in testing for liver disorders.

Verification of a computer-aided replica technique for evaluating prosthesis adaptation using statistical agreement analysis

PURPOSE The purpose of this study was to evaluate the reliability of computer-aided replica technique (CART) by calculating its agreement with the replica technique (RT), using statistical agreement analysis. MATERIALS AND METHODS A prepared metal die and a metal crown were fabricated. The gap between the restoration and abutment was replicated using silicone indicator paste (n = 25). Gap measurements differed in the control (RT) and experimental (CART) groups. In the RT group, the silicone replica was manually sectioned, and the marginal and occlusal gaps were measured using a microscope. In the CART group, the gap was digitized using optical scanning and image superimposition, and the gaps were measured using a software program. The agreement between the measurement techniques was evaluated by using the 95% Bland-Altman limits of agreement and concordance correlation coefficients (CCC). The least acceptable CCC was 0.90. RESULTS The RT and CART groups showed linear association, with a strong positive correlation in gap measurements, but without significant differences. The 95% limits of agreement between the paired gap measurements were 3.84% and 7.08% of the mean. The lower 95% confidence limits of CCC were 0.9676 and 0.9188 for the marginal and occlusal gap measurements, respectively, and the values were greater than the allowed limit. CONCLUSION The CART is a reliable digital approach for evaluating the fit accuracy of fixed dental prostheses.

Effects of image and education on the precision of the measurement method for evaluating prosthesis misfit

Statement of problem: Various methods for measuring prosthesis misfit have been suggested. Although the cross‐sectional images between the crown and abutment are used to evaluate the misfit, the effects of the image and the observers knowledge on the precision of measurement are unknown. Purpose: The purpose of this in vitro study was to investigate the effects of the image and of education on the precision of prosthesis misfit measurement methods using inter‐ and intraobserver variability analyses. Material and methods: The cross‐sectional images in the margin were obtained using the computer‐aided replica technique (CART), silicone replica technique (RT), and sectioning technique (ST). Twenty‐five observers measured the absolute marginal discrepancy values in the images obtained from each group 4 times at an interval of 2 weeks; the observers went through different education sessions regarding the selection of the measurement points. The precision of measurement was determined and compared using the likelihood‐ratio test statistic (&agr;=.05) and the intraclass correlation coefficient with the linear mixed‐effects model. Results: The CART group exhibited the smallest deviations in the measurement variations, followed by the ST and RT groups (P<.001). Additional education on misfit measurements generally decreased the deviation values in all the groups, but without any significant differences. Conclusions: The cross‐sectional image obtained from the measurement methods and education on the evaluation method affected the precision of the prosthesis misfit measurement. Digital methods might be a useful tool to significantly enhance the precision of the measurements.

The Analysis of Roll Call Data from the 18th Korean National Assembly: A Bayesian Approach

We apply a Bayesian estimation procedure to the analysis of roll call voting records on 2,389 bills processed during the 18th Korean National Assembly. The analysis of roll calls yields useful tools for to combining the measurement of legislative preference with the models of legislative behavior. The current Bayesian procedure is extremely exible, applicable to any legislative setting, irrespective of the extremism of the legislator``s voting history or the number of roll calls available for analysis. It can be applied to any legislative settings, providing a useful solution to many statistical problems inherent in the analysis of roll call voting records. We _rst estimate the ideal points of all members of the 18th National Assembly and their con_dence intervals. Subsequently, using the estimated ideal points, we examine the factional disparity within each major party using the estimated ideal points. Our results clearly suggest that there exists a meaningful ideological spectrum within each party. We also show how the Bayesian procedure can easily be extended to accommodate theoretically interesting theoretical models of legislative behavior. More speci_cally, we demonstrate how the estimated posterior probabilities can be used for identifying pivotal legislators.