Cohort explorer
Review eligibility rules, attrition, baseline characteristics, missingness and follow-up together. Export the selected cohort definition with each summary.
Useful for: feasibility discussions and reviewing analysis populations.
EXPERTISE / 03
Focused Python and R/Shiny application development for statistical programming and RWE teams, scoped around your workflow and agreed with your statistical leads.
THE APPROACH
Agree the question, context, constraints and the people who need the result.
Work in agreed milestones with visible assumptions and proportionate quality checks.
Leave you with usable outputs, documentation and a clear account of remaining limitations.
APPLICATIONS FOR STATISTICAL PROGRAMMING
Focused Python and R/Shiny applications, designed with your statisticians and programming team around a defined workflow.
Review eligibility rules, attrition, baseline characteristics, missingness and follow-up together. Export the selected cohort definition with each summary.
Useful for: feasibility discussions and reviewing analysis populations.
Compare dataset versions, highlight metadata changes, review SAS logs and track discrepancies against a specification.
Useful for: repeatable QC and focused vendor reviews.
Run agreed descriptive summaries and reporting templates through controlled inputs, with visible denominators, data versions and downloadable outputs.
Useful for: consistent recurring analyses and transparent handover.
Example build scopes for discussion. Each application needs agreed calculations, validation and access controls before operational use.
DESIGNED WITH STATISTICS
I can help turn a recurring analysis process into a focused Python or R/Shiny application: agree the question with your statistical lead, define the calculations, build a reviewable prototype and document the route into your organisation’s environment.
POTENTIAL BUILDS
Compare two versions of a dataset or output specification and map changed definitions to affected derivations, programs and TLFs. Reviewers confirm the dependencies; the app produces a targeted revalidation checklist. A first prototype could use synthetic specifications and a manually verified dependency map.
Check structured table exports against agreed population counts, subgroup definitions and footnote rules. Highlight inconsistent N values or missing explanation text for programmer review. Start with a small template family and known test discrepancies, rather than attempting to interpret every possible report.
Attach the cohort definition, parameter choices, code revision, data snapshot reference and reviewer decision to each output run. When a result changes, show which inputs changed alongside it. This could make repeated RWE feasibility discussions easier to reproduce without storing patient-level records in the ledger.
Proposed application concepts, not delivered client case studies. Feasibility, scope and acceptance criteria would be agreed with your team.
Technology options include R Shiny, Shiny for Python and Streamlit, selected for the existing code, users and hosting environment. The analytical code should be testable independently of the interface. A scoped handover can include versioned source, agreed test cases, a dependency record, documented permissions and output traceability.
My background includes R/Shiny and Spotfire applications, software prototypes, and support for RWE reporting standards and automation. Python applications are an additional build option for agreed projects; the examples here describe potential engagements, rather than named client deployments.
OHDSI’s cohort characterization workflow illustrates baseline and follow-up summaries. Shiny for Python supports interactive analytical applications, while Streamlit’s architecture guidance explains its execution and state model. The FDA guidance on electronic health records and claims data is relevant when assessing data for regulatory use.
Sources reviewed 21 September 2026. These are independent application concepts, not reproductions of a former employer’s systems.
A CONVERSATION IS A GOOD START
Tell me about the challenge. Let’s define a practical way forward.