 Hi, my name is Mette Benson. I'm a statistician at Novo Nordisk and I'll present the in-and-compare package that three colleagues and I have been developing to support the peer programming process in clinical studies at Novo Nordisk. This is a disclaimer for you to note. A little background information of why we need the functionality of this package. To those of you who do not know, Novo Nordisk is a Danish pharmaceutical company developing pharmaceutical products, especially within diabetes care, but also in other areas as hemophilia and growth hormone disorders. The clinical development process is long and it consists of different phases. The drug discovery phase where we discover and test new candidate drugs in the laboratory and also in animal studies. The conduct of clinical trials to investigate the effect and possible side effects in patients and finally submission of the results to the health authorities to get the approval of the new product. Based on the data collected in clinical studies, we write a clinical study report presenting and interpreting the results and for that purpose we create a lot of tables, figures and listings, TFLs, to ensure a high level of quality of this. Novo Nordisk has defined some procedures regarding the programming and reviewing of the TFLs. Accordingly, key results must be what we have here called peer programmed. The peer programming process at Novo Nordisk involves two persons, the programmer and the reviewer and they are both working on the same task which could for instance be statistical analysis based on key data from a clinical study. The programmer develops a program most often in SAS, but it could also be an R. The program typically produces the data frame with the results of the statistical analysis and also some tables or plots that are to be included in the clinical study report that presents the results. The reviewer creates a peer program that reviews or validates the programmer's work and to avoid being influenced by the programmer's code the reviewer should not read this until after the preparation of the peer program. The peer program typically produces a data frame that should then be compared with the data frame from the primary program and for that you can use the NN compare packets. The actual comparison of the two data frames is done with the compare DF function from the Arsenal packets. The two data frames do not have to be sorted in the same order if you as here specify the buy argument with the key variables. The X function is then called on this summary on this summary object of the comparisons which creates a report with the results of the comparison. The report can be in various formats and furthermore a summary report is created summarizing all comparisons made within the given study. This is a summary report that shows where the differences are detected in other number of columns rows or in any of the compared values of the two data frames and furthermore it contains links to the detailed summary reports. And here's an example of a detailed summary report in this you can see that there are differences detected in the number of columns in the two data frames and furthermore it shows that there are differences in detected in the E variable and also some details about these differences. The NN compare package also makes it possible to compare PND files with the PND compare function. This is used to discover changes in plots that can have enduring for instance finalization of data sets between we have presented the results to management and writing the final clinical study report. Two PND files are compared pixel by pixel and any changes are marked with a red box as depicted here. Time stamp changes are okay and they are marked with a green box. We have some ideas for further development of this package. It could be useful for us to be able to compare more than one pair of data frames within a single function call. Furthermore, we would like to implement an overview of results of several PND file comparisons and then finally we could consider to implement functionality to be able to compare other file types also. The package is available at Github. I thank you for listening in and please let us know if you have any questions, comments to this.