Monday, May 23, 2005
PharmaSUG Monday
PharmaSUG kicked off on Monday morning under the Arizona sun. I checked out Dana Rafiee’s hands on workshop entitled “Get Experience with New SAS Drug Development Software”. There were many features for SDD so only a subset was demonstrated during the workshop. From the client’s perspective, it is rather simple in that all that was needed was a web browser. However, the architecture was rather complex including a web server, along with an application server. There were plenty of components such as the implementation of WebDAV which creates libnames and remote lettered drives on the server. This allows programs to on the client machine to access data on the server.
Some of the features and capabilities that were covered for SDDS included: esignatures for security, all files organized and stored in folders, importing and exporting data, and version control with audit trail. Dana described that the goal for the workshop was to have users log onto SDD and use the data explorer. The users then wrote and executed SAS programs and applied some data importing and exporting. This is only a subset of the capabilities but it was what was possible in the short work shop.
I saw a presentation by Mr. Benze entitled “Risk Based Approach to SAS Program Validation” that made some good points. It was noted that if your SAS programs and computer system is not validated, problems and errors found can cause delays that can be costly for the sponsor. It also creates credibility issues that are difficult to quantify. There were references to Guidelines that strongly recommends that data being analyzed should be validated. The amount of effort that goes into validating a SAS program depends on the risk assessment that the program has. This can range from a simple peer review to a more formal requirements and test plan. It was also important for the SAS programs to be controlled to ensure completeness and accuracy.
I talked to Vincent DelGobbo who is a Senior Systems Developer in the Web Tools Group at SAS Institute. He demonstrated stored processes which is part of the Integrated Technology package. This allows SAS programs to be executed via a Java Servelet in a similar way to how SAS/IntrNet utilizes CGI to execute SAS programs. There seem to be more tools however since it also generates HTML front ends for you as a template compared to SAS/IntrNet which requires you to create all your own custom interfaces. It does this through the integrated technology themes using cascading style sheets to standardize the way front ends are generated.
I got a chance to catch up with Thomas Cox who was part of the XML Integration Team. There are new developments for PROC CDISC which has been in production since SAS 8.2. The data standards established by CDISC are constantly being updated so PROC CDISC is keeping up with the changes. What was new to me was the CDISC Viewer which is only now in beta. It looked a little like the SAS viewer that is available for PC SAS but it viewed define.xml files. This will be available as a free download from the SAS website. I was very impressed with its ability to have hyperlinks that exist from the columns from the define.xml file being viewed. If the meta tags in the define.xml has the proper HREF meta tags, it will display link to an external file similar to how a hyperlink in an HTML file. This is definitely something I plan to download and check out.
The PROC CDISC has support for the ODM version 1.2. However, it appears to have limited support for SDTM and SDS. Thomas explains to me that the XML expressions and schemas has not yet been defined fully for SDTM and SDS and therefore the PROC CDISC can not yet implement these features. SAS is putting efforts into keeping on the edge of all the standards that is coming out of CDISC.
I met Sara Holshouser who was one of the developers for SAS Drug Development software. I was intrigued by some of the new features which I saw from Dana in the morning. The way it organizes your SAS programs and data is similar to how windows explorer works. This is done with a hierarchical folder structure. The data viewer was very useful with features of performing some simple summary statistics. I like how SDD define certain libnames and certain constructs that may server specific as macro variables. SDD handles the management of this by resolving these macro variables to run on its server. The process editor also seems to be more refined from the earlier versions I saw from last year. It is now at version 3.0 with 3.1 in the near future. The process editor is similar to the display manager editor with color coding and all the nice features of accessing online help in case you forget the syntax of a particular PROC. There are plenty of features of SDD which I have not fully get a chance to see so I will probably try to catch up with Sara again tomorrow to learn more.
Some of the features and capabilities that were covered for SDDS included: esignatures for security, all files organized and stored in folders, importing and exporting data, and version control with audit trail. Dana described that the goal for the workshop was to have users log onto SDD and use the data explorer. The users then wrote and executed SAS programs and applied some data importing and exporting. This is only a subset of the capabilities but it was what was possible in the short work shop.
I saw a presentation by Mr. Benze entitled “Risk Based Approach to SAS Program Validation” that made some good points. It was noted that if your SAS programs and computer system is not validated, problems and errors found can cause delays that can be costly for the sponsor. It also creates credibility issues that are difficult to quantify. There were references to Guidelines that strongly recommends that data being analyzed should be validated. The amount of effort that goes into validating a SAS program depends on the risk assessment that the program has. This can range from a simple peer review to a more formal requirements and test plan. It was also important for the SAS programs to be controlled to ensure completeness and accuracy.
I talked to Vincent DelGobbo who is a Senior Systems Developer in the Web Tools Group at SAS Institute. He demonstrated stored processes which is part of the Integrated Technology package. This allows SAS programs to be executed via a Java Servelet in a similar way to how SAS/IntrNet utilizes CGI to execute SAS programs. There seem to be more tools however since it also generates HTML front ends for you as a template compared to SAS/IntrNet which requires you to create all your own custom interfaces. It does this through the integrated technology themes using cascading style sheets to standardize the way front ends are generated.
I got a chance to catch up with Thomas Cox who was part of the XML Integration Team. There are new developments for PROC CDISC which has been in production since SAS 8.2. The data standards established by CDISC are constantly being updated so PROC CDISC is keeping up with the changes. What was new to me was the CDISC Viewer which is only now in beta. It looked a little like the SAS viewer that is available for PC SAS but it viewed define.xml files. This will be available as a free download from the SAS website. I was very impressed with its ability to have hyperlinks that exist from the columns from the define.xml file being viewed. If the meta tags in the define.xml has the proper HREF meta tags, it will display link to an external file similar to how a hyperlink in an HTML file. This is definitely something I plan to download and check out.
The PROC CDISC has support for the ODM version 1.2. However, it appears to have limited support for SDTM and SDS. Thomas explains to me that the XML expressions and schemas has not yet been defined fully for SDTM and SDS and therefore the PROC CDISC can not yet implement these features. SAS is putting efforts into keeping on the edge of all the standards that is coming out of CDISC.
I met Sara Holshouser who was one of the developers for SAS Drug Development software. I was intrigued by some of the new features which I saw from Dana in the morning. The way it organizes your SAS programs and data is similar to how windows explorer works. This is done with a hierarchical folder structure. The data viewer was very useful with features of performing some simple summary statistics. I like how SDD define certain libnames and certain constructs that may server specific as macro variables. SDD handles the management of this by resolving these macro variables to run on its server. The process editor also seems to be more refined from the earlier versions I saw from last year. It is now at version 3.0 with 3.1 in the near future. The process editor is similar to the display manager editor with color coding and all the nice features of accessing online help in case you forget the syntax of a particular PROC. There are plenty of features of SDD which I have not fully get a chance to see so I will probably try to catch up with Sara again tomorrow to learn more.
# posted by Sy Truong @ 11:39 PM 
