bio-clinical-biostatistics-cdisc-data

What it does

Reads, validates, and prepares CDISC SDTM and ADaM clinical trial data for analysis. Handles SDTM domain joins (DM, AE, EX, VS, LB, DS), ADaM architecture (ADSL, BDS, OCCDS, ADTTE) with traceability, treatment-emergent AE conventions, baseline derivation, SUPPQUAL/NSV handling, Define-XML 2.1, and Pinnacle 21 / CORE validation. Recommended when working with clinical trial datasets in CDISC SDTM/ADaM format, preparing analysis-ready data, or validating for regulatory submission.

How it works

• Specifies loading methods for SDTM/ADaM data: Python-based approaches using pyreadstat.read_xport() and pandas merges, and R equivalents for metadata handling (haven.read_xpt, admiral, Pinnacle21/CORE).
• Demonstrates how to load and merge SDTM domains into ADaM structures while preserving subject/event-level relationships and traceability.
• Provides concrete examples: loading datasets, computing indicators (e.g., HAD_SERIOUS_AE), and performing left joins to preserve randomized subjects.
• Documents aggregation strategies and how to aggregate event-level data before merging with DM to avoid row inflation.
• Describes how to derive ADaM structures: ADSL (one row per subject), BDS (one row per subject/parameter/analysis-timepoint/flag) with required fields (USUBJID, STUDYID, PARAM, AVAL, ABLFL, etc.), and OCCDS/ADTT/EH specifics for TEAE and time-to-event analyses.
• Includes executable code snippets showing how to compute ADLB from LB, derive ADY, BASE, CHG, PCHG, and how to build baselines.
• Covers the ADaM CNSR convention for ADTTE and how to map CNSR to standard event indicators for survival analyses.
• Emphasizes Define-XML 2.1 as metadata requirements for ADaM datasets and traceability.

When to use it

• Use when working with clinical trial datasets in CDISC SDTM/ADaM format, preparing analysis-ready data, or validating for regulatory submission.
• Use when requiring site-level and subject-level traceability from SDTM to ADaM, and when performing TEAE conventions, baseline derivation, or Define-XML/CORE validation steps.

What it can touch

• Python: pyreadstat, pandas (pd.read_sas, pd.merge); specific function calls listed in examples: pyreadstat.read_xport(), pd.read_sas(), pd.merge().
• R (conceptual equivalents): haven.read_xpt(), admiral, Pinnacle21/CORE for validation.
• The skill folder ships executable scripts (no explicit external tools beyond the listed ones).

Caveats

• Compatibility notes reference specific library versions (pyreadstat 1.2+, pandas 2.1+, numpy 1.26+; ADaMIG 1.3, OCCDS 1.1, Define-XML 2.1, Dataset-JSON v1.1) and reliance on regulatory guidance (FDA, ICH E2A/E9, etc.).
• Warns about ADaM CNSR convention differences between R/Python survival packages and CDISC CNSR expectations; requires conversion when passing ADTTE results to analysis tooling.
• License indicated as NOASSERTION; explicit licensing terms are not provided in the snippet.