Information Architect | Clinical Data Standards · Semantic Web · AI. Based in the Göteborg area, Sweden.
Information architect with 45 years in the field and 25 of those in clinical trial data and metadata.
I am retired but not done. There is so much interesting work that is now possible with AI, and I want to continue contributing — but selectively, with the right people, on problems that matter. Problems that now can be addressed by using AI to apply sound information architecture principles on existing models and vocabulary standards - and on the data itself, when it actually uses them.
Work
CDISC for AIExplorative work on the behaviour of the CDISC standards, with machine-actionable reference files built from SDTM CT, NCIt and COSMoS.
usdm-rdfThe CDISC Unified Study Definitions Model (USDM) v4 rendered as RDF/OWL, generated mechanically from the published model. Draft.
usdm-overlayReadings of USDM v4 on top of usdm-rdf: what each attribute and class actually holds, with the evidence for each reading, and findings about the model, each with a test. Draft.
cosmos-rdfCDISC COSMoS (Biomedical Concepts and SDTM Dataset Specializations) rendered as RDF/OWL, as published. Early, exploratory work.
soa2usdmSchedules of Activities extracted with AI from protocols published on ClinicalTrials.gov, into layered JSON where every cell traces back to its source page, with review pages for human checking. Two short videos and the schemas on one page introduce it. Stating the result in USDM v4 is work in progress. The extracted collections and the pipeline code are kept in separate repos.