Biostatistics

ADaM-based Table/Listing/Figure pseudocode for every early-phase design (parallel, crossover, single-sequence, SAD/MAD), in both SAS and R (pharmaverse: dplyr + rtables + ggplot2 + emmeans). Geometric PK on the log scale, participants-not-rows, period denominators.

A Quarto + on-prem small-model skeleton that auto-drafts TLF narrative text. The model never sources a number — deterministic facts feed a number-validated draft, behind a human gate. The self-test rejects hallucinated numbers.

The complete Phase-1 clin-pharm deliverables pipeline — 136 deliverables from protocol to CSR — as a register, a reference doc, and a visual map, with a worked PK example.

A public-calibrated synthetic early-phase study running all the tools end-to-end — 8 ADaM domains, a 1-compartment PK + NCA, AE / vitals / labs / ECG, the TLFs, and the SLM — with a Monte-Carlo calibration check.

Four independent Streamlit apps for clinical eCRF extracts, each a four-tier chain with its own audit log: MedDRA-style auto-coding of adverse-event verbatims, PII/PHI flagging and redaction, participant safety and discontinuation risk scoring, and anomaly and drift detection. Labeled [SI] stand-ins run until trained runtimes are installed. The public demos run in a locked hosted mode on synthetic data only (99-prefixed placeholder codes, not MedDRA), with a 244-check QC harness in CI.

Seven named gates that make a clinical-biometrics deliverable delivery-confident — reproducibility, numeric-provenance (numbers from validated tools, never an LLM), CDISC conformance & integrity, SAS↔R double-programming parity, adversarial QC, regulatory/reporting standards, and AI-use governance. The checks are runnable and dogfooded on this repo's own libraries + synthetic data, and emit an evidence ledger.

The biostatistics-AI training walkthrough — lessons, worked examples, a quiz, and a completion certificate. Built to be taken in multiple modalities.

Every library, pipeline, and worked example has a narrated walkthrough (true headless-Chrome screencasts for the live tools). The new-hire path puts them in the order to learn.