metamodeling forms a study period report Denise Warzel, NCICBIIT Jim Davies, Charles Crichton, Andrew Tsui, James Welch, Steve Harris* Department of Computer Science, Oxford Daniel Abler, CERN

agenda why might we need one what instances already exist what are their common features what one might look like what would be its scope where do we go next

motivation specification – I want this data to be collected interoperability – we should be collecting this data – we have collected this kind of data before documentation – this is how this data was collected – no, this is what this data means

motivation … discovery and reuse – of data was any data like this collected? – of form designs can I save some time here? is there some way of collecting this kind of data that is validated? model driven software engineering – I need to guarantee that my system collects the data I have described – I need bespoke functionality, not bespoke software – I want my software to be more general

examples – in paper euro QoL TCGA follow-up form (NCI, partial specification) business process templates

examples - electronic open clinica/redcap slot extensions in EAP software caDSR CDISC ODM HL7 CDA DDI 3.x

Field staticfield runtimecontrol logic presentatio n Exception handlingre-useUser managementother Field definitions (Questions, Value domain) Selection of items from value domain (one, multiple, number) arbitrary field annotation data constraints/ validation pre- population of data fields skip logic execution order including graphics*/ multimedia ** for validation can value pairs be specified by look-up, e.g. web- service? elements are versionable Form roles* / fine grained user access control** attributing users to set of user configurabl e form level metadata audit recordsignature all have some some way of expressing multiplicity of fields//tables, most have multilanguage support, 'mandatory' flag for fields. all have grouping none of the models allows to specify accessibility options like voice or braille none of the models provides discrepancy manageme nt that is configurabl e through model all form elements are referencabl e, at least internally none has (clearly defined) form roles or fine grained user access control,, not possible to attribute users to different sets Documenta tion CDISC-ODMy??* y*/n**/?** *?*/?**yy*"/y**?*/n**ysomey?*/n**nnyy DDIy??* ?*/?**/?** *?*/?**y?*/n**?*/y**nny?*/n**?n?? eDCIyn y*/n**/n** *n*/n**ny*/n** nnnn*/n**nnnn Generation OpenClinicayyn y*/?**/n** *y*/n**yy*"/n**n*/n**ynn nnnn RedCap????????*/?**??????? CancerGridynn n*/n**/n** *n*/n**yn*^/n**n*/n**nnn nnnn caBig Formsynn n*/n**/n** *n*/n**nn*^/n**n*/n**nny nnnn * = provide some annotation capabilitiy, e.g. concept reference, external code list; but not arbitrary * = for single field * = fixed value * = form elements within forms ** = computed across fields ** = computed from fields ** = forms within study

metamodel elements identification static structure and mapping validation control logic/flow

metamodel elements - structure questions – typing – instruction text – their relationship to data elements sections – their relationships to other sections on other forms the container – essential annotation of scope

metamodel elements - validation field-level data checking logic – narrows typing information from ISO11179 – account for dependencies (scope?) default values

metamodel elements - flow ordering paging skip logic to what extent is style in scope?

administered item terminology data elements value domains models

17

scope not a data model or a workflow model – although clearly related not an implementation – although it has to encompass XForms, webforms, JSF, ASP.NET forms … maybe has elements of a meta-standard – may specify constraints on DSLs for validation/flow control/skip logic/workflow

scope … will describe how to specialise ISO/IEC11179 data elements into questions will specialise ISO/IEC19763 – is it simply another component standard of the family will reuse administration modeling from this workgroup

next steps approval to begin turning the research into a standard advice on how the standard is structured and presented – abstract or concrete metamodel? – multipart? – specification + guidance notes?