Graphics Graphics
Transcription
Graphics Graphics
Graphics, Graphics Everywhere and Not a Plot to Print! An overview of the current graphics landscape within Novartis Pharma Development Phuse Single Day Event, June 2012 Stephen Hart, Ulrich Knappich The opinions expressed in this presentation and on the following slides are solely those of the presenter and not necessarily those of Novartis. Novartis does not guarantee the accuracy or reliability of the information provided herein Background § The general use and quality of interactive graphical interfaces is ever increasing with new technologies and software e.g. smart phones, tablets, etc. § Within Novartis Pharma development new technologies are consistently enhancing the way we work e.g. integration of iPad’s for field monitors § Within high caliber scientific publications the amount and quality of graphics seems to be high e.g. Lancet § Yet high quality graphics continue to form an important but only minor part of dossiers submitted to pharmaceutical regulators 2 – Graphics, Graphics everywhere – Phuse SDE Introduction § We are currently analysing the technologies available in Novartis to create our clinical graphics. This is with a view to: Ø Increasing the number of graphics being produced for all types of analysis and reporting activities Ø Rationalizing the products we use Ø Establishing a strategy for which products to use when 3 – Graphics, Graphics everywhere – Phuse SDE Introduction This presentation briefly explores: § Novartis and external landscapes § The pros and cons of the products we at Novartis currently use § A potential change in the way we may produce and view production of graphics within our environment Ø Focus on two technologies recently added to our portfolio: Ø SAS V9.3 ODS Graphics Designer and Tibco Spotfire Clinical Graphics (TSCG). § This is not a sales pitch for any product mentioned!!!! 4 | <Scope-Seek/Planning/Execution> | Project Charter | Project Name | Author | Date | Business Use Only Why Do We Need Graphics? § Question: What have the following items have in common? Ø News report comparing government finances in the Eurozone Ø TV ad for a new toothpaste Ø Self assembly flat-pack cupboard § Answer: They would typically all be presented with an illustration, picture or graphic § Showing a graphic is an easy way to present information in a quick and clear manner § In our industry, increased use of graphics could cut down on internal and external review time, getting drugs to market more quickly J “A Picture Paints a Thousand Words” 5 – Graphics, Graphics everywhere – Phuse SDE Why Do The Dossiers We Submit Contain So Few Graphs? § Clinical data is bulky and complex, not always easy see relationships. So why do we in the pharmaceutical industry not utilize graphics more? Ø Not a formal regulatory requirement? Ø Graphics only needed for presentations and publications? Ø Not enough ”Bang for our Bucks”? Investment in time and money to build/develop an efficient graph application too large? Ø Producing a graph is regarded as a resource intensive activity? Ø Easier to display the numbers (tables produced anyway)? Ø Lack of time to stop and think of new and better ways to display data? Other priority work to attend to? 6 – Graphics, Graphics everywhere – Phuse SDE Why Do The Dossiers We Submit Contain So Few Graphs? Ø When testing ideas, no software tools available to visualize the data presented in different graphical formats. Ø Difficult to determine which graph type better displays certain data e.g. continuous, categorical, nominal, ordinal, distribution of variable Ø Are we always able to keep up with the pace of change? Do we really know what’s around and possible? Ø Or do we just follow the previous set of reports that were produced for a similar activity? 7 – Graphics, Graphics everywhere – Phuse SDE Evolution of the Application Landscape § Did we always have many tools that could easily create high quality graphics efficiently? Do we have these today? § The first version of SAS which I used in 1999 was SAS 6.12 § Graphics production using SAS was technically challenging! § The technology didn’t allow crisp high quality graphics we see today. Device drivers and multiple software were not compatible. Multiple Operating systems needed to get final product e.g. Unix, Linux, Windows, etc. § How could we ensure that this process was reproducible and auditable in line with external regulation? § Many other graphics solutions have become available 8 – Graphics, Graphics everywhere – Phuse SDE Evolution of the Application Landscape § Are we now at a point in time where we can look at producing high quality graphics more quickly and easily? Ø Prepared templates, and code generators mean time to develop a graph can be reduced. Ø Solutions integrated into our landscape allow reproducibility and auditability of graph programming. Ø Code sharing or prototypes of code that do 80-90% of the job e.g. CTSPedia, mean that a wide range of graphics can be produced in a reduced time. 9 | <Scope-Seek/Planning/Execution> | Project Charter | Project Name | Author | Date | Business Use Only Analysis of Current Novartis Landscape § My experience covers both “End-user”, “SAS Developer” and IT business analyst roles § I have seen and viewed the landscape from a few angles § After thinking about the amount number of graphics tools we currently have available to us many questions came to me: Ø What tool should be used for what purpose? Ø How does a user know which tool should be used when? Ø How long does it take to create a report for each tool? Ø Which tools are used by which groups? 10 – Graphics, Graphics everywhere – Phuse SDE Analysis of Current Novartis Landscape § Recently I had the opportunity to see a demo from SAS on “High performance analytics” Ø SAS visual analytics applications used mainly for marketing, forecasting and commercial purposes Ø Exceptional amount of “Big Data” being crunched and displayed very quickly Ø One billion observations analyzed and displayed in minutes § We need to modify our approach to use these technologies Ø Difficult to use “out-of-the-box” tools in our regulated environment 11 – Graphics, Graphics everywhere – Phuse SDE Reports, Paper, Action! / 88 numbers but what's the story.... § This small snapshot of the first page of a typical set of summary information –it contains 88 numbers § In large outcome trials a single AE incidence table may cover more than 100 pages § But what does it tell us? § Is it really possible to review this information in its fullest form? 12 – Graphics, Graphics everywhere – Phuse SDE 13 14 Application Vrs Characteristics Matrix (Novartis) Characteristics Application JMP SAS R S-plus Spotfire TSCG Technical knowledge & complexity needed to create a basic graphic? No. (iPad touch version) Yes / No –ODS GD Yes, (progra m) As R Data must be prepared Yes / No Can be purely Graphical Produces Interactive displays i.e. graphic that allow users to point on values. Yes, brushing, is dynamic Yes – Can be in html No No Yes No Purpose built templates included (OOTB) No. JMP/ Clinical Only Yes (ODS GD) No No Yes Yes Can be audit trailed? Should it be used for regulator work? No Yes Maybe ~valida tion? Yes No Yes High quality resolution graphics available on Novartis systems?* Yes Yes / No – ODS GD Yes Yes Yes Yes Is it compatible with all OS? e.g. windows, Unix, Linux, iPad, etc. No Yes Yes Yes No Yes / No Exploration & marketing Running predefin ed Graphics Free. modern langua ge data minip facil. BI like displays, can be planned ahead, Excellent for Medical Review Templates & Config. displays What area does it provide maximum benefit currently in Novartis? 15 – Graphics, Graphics everywhere – Phuse SDE TSCG, SAS 9.3 ODS GD and GPS II Overview § GPSII is the Novartis environment created to support clinical analysis and reporting in terms of traceability and auditability Ø Development and final “Production” execution done in a secure environment Ø Ensures compliance to 21 CFR part 11 guidelines § Trying to get the best of both worlds!! Ø Increase the usage of “out of the box” configurable solutions Ø Use existing templates which support the standardization effort Ø Easy to modify (configure) templates via GUI Ø Integrate graphics applications into our validated environment Ø Code generator for secure execution within GPS II 16 |– Graphics, Graphics everywhere – Phuse SDE TSCG, SAS 9.3 ODS GD and GPS II Overview (Cont.) § What do TSCG and SAS ODS GD give us “out of the box”? Ø Defined set of fixed templates – TSCG: ~25/30 - SAS:~35-40 § Both solutions allow different user groups to use the tool in different ways Ø Non-technical groups can use the GUI (point and click) as long as they have access to the data Ø Information is saved in graphics configuration files or programs Ø Technical groups can then use these to create code and execute in secure way 17 – Graphics, Graphics everywhere – Phuse SDE TSCG, SAS 9.3 ODS GD and GPS II Overview (Cont.) 18 – Graphics, Graphics everywhere – Phuse SDE User View - Tibco Spotfire Clinical Graphics (TSCG). § Quick wins: Ø Quick and easy to get a first draft and reproduce via SAS macro. Ø Works together with SAS macros and SAS data Ø Works in GPS environment. Ø Good quality graph, easy to have a PDF file as an output or other format. Ø Multiple graphs per page is possible (e.g. Trellis). § Challenges: Ø Sometimes you need to add some technical imagination to get final results you require e.g. annotations, information in margins Ø Can be difficult to get what you want if it is not produced by “out of the box” template. Certain amount of “black-box” technology in terms of the SAS macro code Ø If you generate S-Plus code then you need to be able program in S-Plus to modify. 19 – Graphics, Graphics everywhere – Phuse SDE User View – SAS ODS Graphics Designer § Quick wins: Ø Available on our GPS environment with SAS V9.3 Ø Straight forward to produce a graphic very quickly Ø Code generated automatically (no black-box) Ø Save straight away in a SAS program and modify as needed Ø Easy to create SAS program and reproduce a graph. Ø Drag and drop of multiple graphs possible § Challenges: Ø Still working on making the selection of remote libraries possible Ø How do we make this available to user groups who don’t use SAS? – Stand alone version? 20 – Graphics, Graphics everywhere – Phuse SDE Conclusion § Technology alone may not result in more graphs being created. We also need to: Ø Ensure right tools are used for right jobs! Ø Ensure organizational integration is right for the right groups! Ø Support users when trying new approaches § More agreed “Standard” graphics reports should be defined for use in CSRs Ø More guidance on how and when they should / must be used § External enforcement? Ø Should we view summary tables a bit like we view listings, that is, becoming less relevant? Ø Remove certain portion of tables as is proposed with all listings? Ø Allow only tabular reports which add value and information is digestible? 21 – Graphics, Graphics everywhere – Phuse SDE Conclusion (cont.) § Could we get more for our resource if we produced more graphical outputs? § Could internal and external reviews be speeded up if we had rules and regulation on the production of graphical reports? Ø Maybe for internal colleagues it also triggers more ideas J / L § Could we get a clearer definition from vendors of how to integrate their technologies into our regulated environment to allow us to achieve compliance more easily? Ø Integration of new technologies is vital! § Could the quality of what we produce be enhanced further by using graphics also as a QC tool? Ø Our brain is very good in detecting unusual patterns 22 – Graphics, Graphics everywhere – Phuse SDE Questions, Thought’s or Comments? 23 – Graphics, Graphics everywhere – Phuse SDE Back-up1 – Code Produced by SAS ODS GD proc template; define statgraph sgdesign; dynamic _TRT _VISITNUM _TREATMENT; begingraph; entrytitle halign=center 'Type in your title...'; entryfootnote halign=left 'Type in your footnote...'; entrytitle halign=center 'Type in your title...'; entryfootnote halign=left 'Type in your footnote...'; layout lattice / rowdatarange=data columndatarange=data rows=1 rowgutter=10 columngutter=10; layout overlay; boxplot x=_VISITNUM y=_TRT / group=_TREATMENT name='box' boxwidth=0.4 clusterwidth=1.0; endlayout; sidebar / align=bottom spacefill=false; discretelegend 'box' / opaque=true border=true halign=center valign=center displayclipped=true order=rowmajor; endsidebar; endlayout; endgraph; end; run; options orientation=landscape; ods rtf file="v:/hartst2_view/CSTSTEST/CSTSTEST0001/util/demo1.rtf"; proc sgrender data=WORK.MEANT template=sgdesign; dynamic _TRT="TRT" _VISITNUM="VISITNUM" _TREATMENT="TREATMENT"; run; ODS rtf close; 24 – Graphics, Graphics everywhere – Phuse SDE Back-up1 – Macro Call from TSCG %autorun; %u_prerep; ods html file="/vob/CCTEST/CCTESTSTUDY1/report/export/pgm_saf/dot_plot.rtf"; %tscg_titles(mainTitle=1[main 1] 2[main 2] 4[main 4], subTitle=1[sub 1] 3[sub 3], header=2[head 2] 4[head 4], footer=1[foot 1] 3[foot 3] 5[foot 5]); %tscg_Legend(visible=T, legendLocation=RightCenter, numberOfColumns=1, insideLegend=True, title=Test legend); %tscg_creategraph(outputFile=/vob/CCTEST/CCTESTSTUDY1/report/export/pgm_saf/dot_plot.png, graphDoc=/vob/CCTEST/CCTESTSTUDY1/report/pgm_saf/new_box_plot.igd); ods html close; 25 – Graphics, Graphics everywhere – Phuse SDE