 Good afternoon, I'm Cezanne Freeman from the University of Leicester and today I'd like to introduce you to MetaDTA. MetaDTA is an interactive, web-based, or shiny application which conducts meta-analysis of diagnostic test accuracy studies. So diagnostic test accuracy studies are often reported in terms of sensitivity and specificity, and we synthesize them using a bivariate random effects model which jointly synthesizes sensitivity, specificity and takes into account their correlation. The results can often be displayed as either a summary point in ROC space, so we have sensitivity on the y-axis and 1-specificity on the x-axis, and the blue square in the middle represents that summary estimate with pooled sensitivity and specificity, with 95% confidence and predictive regions, or we can plot a summary receiver operating characteristic. So now I'd like to show you how MetaDTA works. So just to say that on the home page if we just scroll down there is a user guide here which can be downloaded and gives lots of useful information on using MetaDTA. The web page itself can be navigated using this top bar across the different pages if we click on to load data. This page gives the option for selecting your own file as well as instructions for how to format that file, or you can choose to use one of the included data sets so that you have a play around with the different options. On the Meta Analysis page we have a number of tabs. First of all, study level outcomes. So this will give us sensitivity and specificity estimates for each of our study as well as weights, and there are both sorting and search functions here which can be useful with larger data sets. On the SROC plot tab we see our plot in ROC space, and this is the default with the summary point in the middle and the confidence and predictive regions. Over here on the left-hand side we have a number of options for customizing that plot including adding the SROC curve itself. We can also display the city's prevalence estimates and the percentage study weights which are displayed by changing the circles for the data points into ellipses so that the width vertically represents sensitivity weights and horizontally represents the specificity weights. I'm just going to remove those options because we can also display sensitivity and specificity 95% confidence intervals. And we can also have a look at quality assessment scores. So quality assessment is supported in MetaDTA from the QuadAS2 tool, of which there are seven domains and we can choose to display each domain individually, where the red here is high risk of bias and green is low risk of bias, or we can have a look at all seven domains at the same time to remove those confidence intervals. And we can see that that's done by changing our data points into these mini pie charts. If we click on the middle of the pie chart and scroll down, then we see the study name, the sensitivity and specificity estimates, and a bigger pie chart where we can see which domains of QuadAS2 have the high and unclear risk of bias. We can also choose to display covariates. So if we were to look at country, the default option here is to add the covariate as a text next to the data point, but we can also have a look at them as colored points or both, which may allow you to identify any specific trends relating to covariates. On the statistics tab, we get the meta-analysis results. So these are the pool estimates of sensitivity and specificity. And down here on the left-hand side, we have options for displaying other statistics, which are also useful for diagnostic test accuracy meta-analysis. On the parameter estimates tab, this is useful if you want to include your meta-analysis in a decision modeling framework. We have parameters for REVMAN, which is useful if you want to construct any of these ROC plots within REVMAN. And then we have the forest plots page for a forest plot, the sensitivity and the specificity. All the plots in meta-DTA can be downloaded as either PNG or PDF files, and all the tables can be downloaded as CSV files. Clicking onto the sensitivity analysis page. All the options within this page, so all the tabs, all the options are the same as the meta-analysis page, but we just have this additional option down the bottom here, where we can choose to exclude studies from our meta-analysis. So I've just excluded two studies where all domains of quadacity were at high or unclear risk of bias. Clicking onto the SROC plot tab. In this page now, the model including all studies is in grey in the background here and in the foreground in blue is sensitivity analysis excluding those two studies. So we can see here there was not much impact. And onto the statistics tab. We can get an additional table here, which gives the estimates just for the individual studies, and you can see them side by side with the estimates from all studies. The rest of the tabs here relate just to the studies that are being included in your sensitivity analysis. And the final tab is prevalence, giving you an opportunity to have a look at how many patients you would expect to be testing positive and having the disease, or we can turn it around the other ways that we look first whether they're diseased and then whether you'd expect them to test positive. There's also an option for the sensitivity analysis, which relates back to the studies that have been excluded on the sensitivity analysis page. And the references show that we've used 22 different R packages here to build this application. So thank you all very much for listening, and I'd be very happy to answer any questions. Thank you.