 Hello, my name is Genguna Ashima. I'd like to thank the Esma Khan of the 2018-3 organizing team for kindly inviting me to speak here today. I'd like to talk about PyMeta package, which is an R package for prediction intervals for random effects meta-analysis. Random effects meta-analysis combine the treatment to affect estimates across these accounting for heterogeneity. One of the objectives of a random effect meta-analysis is to summarize studies and to make statistical references on the abrasive intermittent effect. Quantification and variation of the magnitude of heterogeneity are also very important, because true treatment effects can differ for each study due to differences in patient characteristics follow-up times, rough regiments, and other factors in main cases. However, researchers often interpret results from random effect models in the same manner as those from fixed effect models that assume the true effect does not differ from study to study. So, a prediction interval was proposed. Prediction intervals can be interpreted as a range of the predicted true treatment effect in the new study. Prediction intervals provide useful additional information to confidence intervals, because a prediction interval is a measure of treatment effect that accounts for heterogeneity. It was recommended that a prediction interval should be reported alongside a confidence interval under heterogeneity measures. You can find existing methods in this list. Several methods to construct prediction intervals for random effects meta-analysis has been developed. The method of Higgins et al. is most commonly used. This method could have above coverage for heterogeneity with a number of studies is small, because this prediction interval is based on large sample approximations. So, modified method was proposed. Hartman-Clape, Kenward-Roger, and Bootstrap methods can be used. These predictions intervals have better coverage for heterogeneity with a number of studies is small. Now, I'd like to show you the primate package with examples. This package is easy, load your data, and then pass it to the primer function. The 3-laner website has examples that are described here. This slide shows an example code with a package example data. Code counts are continuous values in this data. Y is the vector of effect size estimates. S e is the vector of parameters of the within study standard errors. B is the number of Bootstrap samples. C is the random number seed. Parallel is the number of threads used in parallel computing. Here, we have a result calculated with the primate package. An estimated result can be summarized by the print method. Both prediction and confidence intervals and summary statistics that are usually reported in the random effect method are provided. An estimated result can also be summarized by a forest plot by the plot method. The forest plot function has arguments to study levels to set the levels for each study. Now, I'd like to show you an example of binary outcomes. The primate package can handle random hex meta analysis with binary outcomes. The convert bin function converts binary outcome data to effect size estimates and resists the standard error vector. This function has an argument type that is a character indicating an outcome measure. Logos range, log risk range, risk difference can be specified. Finally, input data M1, N1, M2, M2, which are integer vectors are converted to effect size estimates and resists the standard errors. Estimated values on the logarithmic scale can be back-transformed to the original scale with the trans-option in print and plot functions. Here, you can see a forest plot for binary outcome. The trans-option can also be used in the plot function. Now, I'd like to show you another feature of the primate package. The primate package has CIMF function for confidence intervals. Many, many methods to construct confidence intervals for random effect meta analysis have been developed. This function supports eight types of confidence intervals listed below. The primate package has TOW squared heart function for heterogeneity variance estimators. Many, many methods to construct heterogeneity variance estimators for random effect meta analysis have also been developed. This function supports 12 types of point estimators and two types of interval estimators. Finally, here is some important additional information. Some functions from the primate package can also be used from the meta package. This slide shows an example code using the meta package. The meta package can also be used to create beautiful graphs. Here, you can see a forest plot using the meta package. You can also use other useful functions that are implemented in the meta package. Future updates will include enhanced graphics capabilities, adding other prediction interval methods, adding other confidence interval methods, heterogeneity variance methods. And I'm also now working on prediction interval methods for meta regression. If you have any other ideas, please let me know. If you have any other ideas, please let me know. I briefly summarized the recent developments in methods of prediction intervals for random effect meta analysis. I introduced the PyMeta package. PyMeta package provides a function to compute a prediction interval and generates a forest plot for the estimated results. Thank you for watching this video. Thank you very much.