pharma-skills

Simulates an independent statistical reviewer auditing a clinical trial submission package (SDTM, ADaM, TLG/TLF, SAP, CSR). Use when the user provides clinical trial datasets, tables, listings, figures, analysis outputs, or submission materials and wants an independent check of correctness, consistency, traceability, or data realism. Reviews denominators, populations, endpoints, multiplicity, safety summaries, cross-layer links from TLF to ADaM to SDTM/source, and whether data looks clinically plausible. Trigger on requests to review TLFs, audit a submission package, verify an analysis, reproduce endpoints, check SDTM/ADaM, assess data quality, or detect fabricated or unrealistic trial data across any therapeutic area.

Derives an ADaM Subject-Level Analysis Dataset (ADSL) using the {admiral} R package and pharmaverse ecosystem. Use when a user needs to create ADSL from SDTM domains, derive standard subject-level variables (treatment dates, disposition, demographics, population flags), or generate QC-ready R code following CDISC ADaM conventions. Requires SDTM input data and an ADaM spec.

Parent skill for the admiral ADaM derivation family. Covers shared conventions used across all admiral child skills: library setup, pipe style, date derivation rules, flag variable conventions, and QC patterns. Route to a child skill for dataset-specific derivation workflows.

Design and simulate clinical trials using the TrialSimulator R package and produce a QC-ready build-order-spine report that pairs each block of code with rationale, parameters, and operating characteristics.

Auto-discover all skills with evals in RConsortium/pharma-skills, benchmark each with vs. without skill using matched isolated sessions, and post scored results to the linked GitHub issue. Use whenever someone says "run benchmarks", "compare skill performance", "eval the skills", or wants to measure whether a skill improves output quality.

Converts one or more GitHub Issues into standardized benchmark data using automated scripts. Use when a user provides an issue number or URL and wants to add it to the evaluation suite.

Derives CDISC SDTM domains from raw clinical (EDC/eCRF) data using the {sdtm.oak} R package. Use when a user needs to map raw study data to SDTM Events (AE, CM, MH), Findings (VS, LB, EG), or Interventions (EX) domains following the sdtm.oak algorithm framework. Produces executable, submission- ready R code with controlled terminology recoding, ISO 8601 date derivation, sequence numbering, and study day calculation.

Derives ADaM Basic Data Structure (BDS) datasets using the {admiral} R package. Initial scope covers ADVS (vital signs) and ADLB (laboratory values). Use when a user needs to create a BDS findings dataset from SDTM domains, derive parameter assignments, baseline values, change from baseline, visit windowing, or analysis flags, following CDISC ADaM conventions. Requires SDTM input data, an ADaM BDS specification, and a completed ADSL.

End-to-end R/pharmaverse workflow to simulate individual patient data (IPD) for a registered clinical trial using a g-formula causal-DAG simulator. Given an NCT ID with posted protocol + results, derives SDTM-style CRFs, builds evidence-based structural causal models, parameterizes them from ClinicalTrials.gov / literature priors, runs forward simulation in R, creates SDTM/ADaM outputs with pharmaverse packages, and iteratively calibrates marginal statistics to the published results without breaking causal identifiability.

Derives an ADaM Adverse Events Analysis Dataset (ADAE) using the {admiral} R package and pharmaverse ecosystem. Use when a user needs to create ADAE from SDTM AE and supporting domains, derive standard adverse event analysis variables (severity, seriousness, treatment-emergent flags, study day variables, baseline flags), or generate QC-ready R code following CDISC ADaM conventions. Requires SDTM input data, ADSL, and an ADaM spec.

Design group sequential clinical trials for survival endpoints (OS, PFS, DFS) with interim analyses, spending functions, boundaries, multiplicity, and event/enrollment prediction. Triggers on: Phase 3 trial design, sample size/events for survival endpoints, alpha spending, group sequential design, interim analysis planning, or enrollment/event timeline prediction for clinical trials.

Generate a combined benchmark analysis for the group-sequential-design skill by reading all benchmark GitHub issues, selecting the latest completed run per issue, and producing a structured three-section report (summary table + overall scorecard + failure pattern analysis). Use this skill whenever the user asks to update the benchmark summary, generate the benchmark analysis, summarize skill vs no-skill results, add failure patterns, or produce `benchmark_analysis_YYYY-MM-DD.md`. Always invoke for any request to compare skill performance across issues over time.

Generate a concise weekly progress summary for the pharma_skills repository. Use this to summarize recent commits, PRs, and issues, then post the update to Slack and save it as a markdown file.