Data conversion
XPT ⇄ CSV Converter
Read and write ASCII SAS XPORT V5, inspect datasets and keep metadata with a CSV schema sidecar.
Local Web Worker · limited V5 subset · independent R fixture checks
Clinical programming.
Research &
practical tools.
I build governed analytics and AI platforms for clinical trials — and lead the teams that run them.
Associate Director, Statistical Programming
Kardigan · PhD, Data Science (2026)
Clinical practice. Open research. Tools you can use.
Free browser tools
Built for the work.
Convert data or build reporting shells. Core processing stays in your browser; outputs need independent review.
Clinical reporting
Mock Shell Generator
Build table, figure and listing shells, review specification bindings, and export drafts or SAS/R scaffolds.
Browser project storage · generated code requires independent review
Governed TFL platform
Live Platform Demo
Run a synthetic study review with an independent QC gate, build an ADaM dataset, or render a clinical figure — real jobs on a governed API, with hash-pinned artifacts and a ZIP evidence bundle for every run.
Anonymous quota 5 runs/day per IP · synthetic / CDISC pilot data only · no account or upload path
Package governance
R Package Validation
Screen R packages with a Tier-1 riskmetric-based assessment, receive a GO/NO-GO decision, and download an evidence-chain report with a hash-chained audit trail for every assessment.
Remote demo on app.jaimeyan.com · anonymous quota 5 runs/day per IP · Tier-2 full validation is a managed service (jaimeyan.com/#contact)
Clinical Data Lab
From question to evidence.
Explore clinical questions through reproducible R and Python, accessible results and independent numerical checks. Every preview comes from an executed analysis.
CDISC Pilot · public test data
How does the outcome change?
Seven connected cases: population, disposition, efficacy, safety, subgroups and time to first dermatologic event.
Explore the Lab →
Figure Library · synthetic teaching data
Make the uncertainty visible.
Six editable R/Python templates with full-size figures, numerical tables and verified exports. Separate fictional data, with required Jaime Yan attribution.
Browse the Figure Library →Selected writing
Ideas, examined.
Clinical R in Practice: The Open-Source Stack in Pharma
A 15-part series on R in pharma: pharmaverse ADaM, ARD tables, risk-based validation, targets pipelines, GxP Shiny, LLM agents — collected into a free bilingual book.
Read the articleThe Clinical SP Bootcamp: A Learning Path from Raw Data to TLF
A systematic path into clinical statistical programming: CDISC fundamentals, modern SCE workflow, and where AI agents fit — every part with runnable takeaways.
The Contradictions Your Validator Can't See
Rule-based SDTM validators structurally miss cross-domain contradictions. SHACL-SPARQL graph constraints plus a deterministic agent layer catch all 20 archetypes.
About
A background in clinical practice
My work spans 9+ years in clinical trials, across cardiovascular and oncology programs. I am the statistical programming lead for the tonlamarsen clinical program at Kardigan — an investigational antisense therapy for uncontrolled hypertension. Previously at Merck, I led statistical programming for LIPFENDRA® (enlicitide / MK-0616) from Phase 2b through the pivotal Phase 3 CORALreef program to FDA approval in July 2026 — the first and only oral PCSK9 inhibitor on the market.
Alongside industry work, I completed a PhD in Data Science at Harrisburg University (2026), researching agentic AI and LLM automation for clinical trial programming. I created py4csr, an open-source Python framework for clinical study reporting.
Leadership
-
Program Lead
Tonlamarsen clinical program, Kardigan
-
NDA Submission Lead
LIPFENDRA® — FDA-approved 2026
-
GenAI Initiatives Lead
Enterprise LLM/RAG platform, Merck
-
eSubmission-Benchmark
Public dataset cited at PharmaSUG 2026
Domains
-
Cardiovascular
PCSK9, hypertension; Phase I–III + CVOT (~14,500 subjects)
-
Oncology
Solid tumors, RECIST 1.1, Phase I–III
-
Regulatory
NDA/BLA, ISE/ISS, eCTD, 21 CFR Part 11
Research
-
Agentic AI
ClinAgent, GxP-Agent architectures
-
Graph-Constrained Validation
CAVE-Onc, PLOS One 2026
-
Code Conversion
SAS ↔ R GenAI pipelines
-
Knowledge Graphs
Neo4j for clinical data
Education
PhD, Data Science — Harrisburg University of Science and Technology
2026 · M.S. Analytics, Harrisburg University (2022, GPA 3.90) · M.S. Applied Mathematics, New Mexico Tech (2017) · B.S. Mechanical Engineering, Jilin University (2015)
Certifications
SAS Certified Advanced Programmer
SAS Certified Base Programmer
9+
Years in clinical trials
30+
Research outputs
2
Peer-reviewed journal papers
13
Conference papers
Publications
Research & Writing
2 peer-reviewed journal papers · 13 conference papers · 2 posters · 15+ preprints & archives
Google Scholar profile →All PhUSE / PharmaSUG / CSS works below were conceived, executed, and written solely by Jaime Yan; on multi-name bylines, additional names appear per conference submission policies.
Peer-Reviewed Journal Articles — Sole Author
PLOS One · 2026-08-14 · doi:10.1371/journal.pone.0350376 · sole author
ClinAgent: AI-Assisted Methodology for Clinical Trial Data Processing and Statistical Programming
Biology Methods and Protocols · 2026 · doi:10.1093/biomethods/bpag032
Conference Papers — PhUSE & PharmaSUG
AI-201
PharmaSUG 2026 · 2026 · doi:10.5281/zenodo.22182922 · with Jason Zhang · companion code
AP-211
PharmaSUG 2026 · 2026 · doi:10.5281/zenodo.22182914 · companion code
OS08
PhUSE US Connect 2025 · 2025 · doi:10.5281/zenodo.22182891 · with Tingting Tian
ET01
PhUSE US Connect 2025 · 2025 · doi:10.5281/zenodo.22182897 · with Tingting Tian
DH03
PhUSE US Connect 2025 · 2025 · doi:10.5281/zenodo.22182899 · with Changhong Shi
ML12
PhUSE US Connect 2025 · 2025 · doi:10.5281/zenodo.22182901 · with Chao Su
AI-239
GenAI Assisted Code Conversion: From SAS to R Standard ADaM Templates
PharmaSUG 2025 · 2025 · doi:10.5281/zenodo.22182918 · with Jeff Cheng, Srinivas Malipeddi, Gurubaran Veeravel, Suhas R. Sanjee
SI-342
PharmaSUG 2025 · 2025 · doi:10.5281/zenodo.22182920 · sole author
IC08
AI-Enhanced Chatbot for Streamlined Clinical Trials Analysis and Document Management
PhUSE US Connect 2024 · 2024 · doi:10.5281/zenodo.22182904 · with Chao Su, Changhong Shi
SI-160
LLM-Enhanced Training Agent for Statistical Programming
PharmaSUG 2024 · 2024 · doi:10.5281/zenodo.22182924 · with Jason Zhang
MM-226
Methodology for Automating TOC Extraction from Word Documents to Excel
PharmaSUG 2024 · 2024 · doi:10.5281/zenodo.22182916 · with Jeetender Chauhan, Madhusudhan Ginnaram, Sarad Nepal
SD-084
A Macro Utility for CDISC Datasets Cross Checking
PharmaSUG 2023 · 2023 · doi:10.5281/zenodo.22182926 · with Chao Su, Changhong Shi
QT-085
Tips to Read In and Output Excel Spreadsheets in SAS
PharmaSUG 2023 · 2023 · doi:10.5281/zenodo.22182912 · with Chao Su, Changhong Shi
Posters — PHUSE/FDA Computational Science Symposium
PP02
JSON Data Generation: Linking Statistical Analysis with Large Language Models
PHUSE/FDA Computational Science Symposium (CSS) 2024 · 2024 · doi:10.5281/zenodo.22182906 · with Changhong Shi
PP20
PHUSE/FDA Computational Science Symposium (CSS) 2023 · 2023 · doi:10.5281/zenodo.22182908 · with Changhong Shi, Chao Su
Selected Preprints & Technical Reports
Preprint (submitted to Pharmaceutical Statistics) · 2026-10-06 · sole author · under review
GxP-Agent: Process-DAG Topology for Reliable Clinical Trial Programming with LLM Agents
arXiv · 2026
arXiv · 2026-05-13 · doi:10.48550/arXiv.2605.13905 · DBLP-indexed
ClinAgent: A Five-Layer Architecture for Autonomous Clinical Trial Statistical Programming
medRxiv · 2026-01-16 · doi:10.64898/2026.01.09.26343542 · preprint of the journal publication
medRxiv · 2025-12-29 · doi:10.64898/2025.12.24.25342988 · with Jason Zhang, Tingting Tian · cited by a Cytel author at PHUSE US Connect 2026
Additional archives: 4 TechRxiv preprints (2025) underlying the PHUSE 2025 papers · Graph-Constrained Skill Loading (JMIR Preprints, 2026) · LogicGraph-HL (OSF, 2026) · 3 Zenodo software/paper archives (2026)
Projects
Selected Work
Flagship Research
·Peer-reviewed, Oxford University Press 2026
ClinAgent
A five-layer architecture for autonomous clinical trial statistical programming. Phase 3 studies typically require 12–24 FTE-months of manual TLF programming; ClinAgent orchestrates LLM agents against CDISC-grounded validation gates to compress that timeline from months to days.
01
Requirements
Protocol & SAP ingestion
02
LLM Planning
Agent orchestration
03
Code Generation
SAS / R output
04
Validated TLFs
CDISC compliant
Platform · Concept Demo
·Company-neutral re-implementation · synthetic data
Governed Clinical Analytics Platform
A 21-service containerized analytics platform for clinical trial data review — one reverse-proxied entry, governance at the platform layer (tiered GxP classification, tamper-evident audit chain, dual-signer e-signature), and zero-egress AI with page-level citations. Explore the architecture and five interactive module demos, re-implemented for the browser with synthetic data.
21 services 100% replay fidelity · 56k records ~96% storage reduction 0 external API calls
Explore the platform demo →
Online Book · EN / 中文 · Free
Modern R in Practice
From Analysis Scripts to Reliable Deliverables — a free 28-chapter bilingual book for R users covering analysis workflows, professional deliverables, package development, and AI/clinical applications, with a companion hands-on lab.
Read the book →Online Book · EN · Free
Clinical R in Practice
The Open-Source Stack Behind Regulated Drug Development — the Clinical R in Practice series collected as a book, expanded with exercises and full case studies: the pharmaverse ADaM toolchain, regulatory-grade tables, risk-based validation, and GxP Shiny.
Read the book →admiral · R · Human-in-the-loop
admiralagent
Trace an ADaM specification through a reviewable Layer IR to deterministic admiral R drafts. Explore validation gates, human revisions, and public CDISC pilot benchmarks.
Explore the interactive demo →Live demo · Synthetic data · No account
Run the Platform Live
Three real contracts on the governed tfl-platform API: review a synthetic study where an independent QC gate can block the report while the evidence stays available, build an ADaM dataset, or render a figure — then download hash-pinned artifacts and a ZIP evidence bundle for every run.
Try the live demo →Open Source · PyPI
py4csr
Python for Clinical Study Reporting — an open-source framework that automates clinical study reporting workflows, published on PyPI with a Zenodo-archived software DOI.
View on PyPI →PharmaSUG 2026 · AI-201
AI-Assisted QC Code Generation
Eliminates QC programming duplication through Claude AI-assisted independent code generation — with companion open-source code and the public eSubmission-Benchmark dataset.
View on GitHub →PharmaSUG 2025 · AI-239
GenAI SAS ↔ R Code Conversion
LLM-powered translation pipeline converting SAS to standard ADaM R templates with validation and CDISC compliance checks — core contribution to an enterprise GenAI initiative.
Read paper →PLOS One 2026 · Sole Author
CAVE-Onc
Graph-constrained agentic validation for cross-domain contradictions in CDISC oncology submissions — knowledge-graph-grounded AI validation.
Read paper →PhUSE 2025 · DH03 / SI-342
Clinical Knowledge Graph
LLM-powered natural language interface to Neo4j graph databases for clinical trial data exploration, synthetic ADaM generation, and relationship discovery.
Read paper →Dash · Shiny · metalite
Enterprise Analytics Platforms
Enterprise data visualization and AI tools for automated clinical trial reporting — interactive DMC reporting on the metalite / forestly / boxly R ecosystem, validated under 21 CFR Part 11.
Request a demo →Learning Center
Learn clinical SP by doing
8 learning lines · 21 interactive lessons · 245 scenes · 45 video companions
Browse the curriculum →The full path from raw EDC export to a submission-ready TLF package — taught as interactive lessons, not lectures. Concept scenes, checkpoint quizzes with graded answers, and hands-on exercises built on a simulated oncology study. Free, no account, progress stays in your browser.
From the NDA statistical programming lead of an FDA-approved product · series archived on Zenodo · companion research cited at PharmaSUG 2026 · AI tutor unlocks by passing quizzes, not by paying
Experience
Professional Experience
May 2026 — Present
Kardigan, Inc. · Princeton, NJ
Associate Director, Statistical Programming
- Statistical programming lead for the tonlamarsen program — investigational antisense therapy for uncontrolled hypertension; KARDINAL Phase 2 published in JACC 2026
- Lead analysis datasets, TLFs, define.xml, and regulatory-ready eSubmission packages for Phase 2 and later-stage trials
- Architected enterprise clinical analytics & AI platforms; established a standardized, reproducible analytics environment under 21 CFR Part 11 / GAMP 5
Nov 2021 — May 2026
Merck & Co., Inc. · Rahway, NJ
Lead Scientist, Statistical Programming
- NDA statistical programming lead for LIPFENDRA® (enlicitide / MK-0616) — first and only oral PCSK9 inhibitor, FDA-approved July 16, 2026; Phase 2b lead (JACC 2023), pivotal Phase 3 CORALreef program (NEJM 2026; CVOT ~14,550 participants)
- Led NDA/BLA eSubmission packages — Define-XML v2.1, reviewer's guides, eCTD, ADRG — in a 21 CFR Part 11 validated environment
- Led GenAI automation initiatives (enterprise LLM/RAG platform); created py4csr, the open-source Python framework for clinical study reporting; CRO vendor oversight; SAS-to-R migration tooling
Jun 2020 — Nov 2021
Regeneron Pharmaceuticals · Basking Ridge, NJ
Senior Statistical Programmer
- SDTM/ADaM development for Phase 1–3 oncology trials (solid tumors, dose escalation); RECIST 1.1 endpoints
- Ad-hoc safety analyses supporting data review and regulatory interactions; external vendor data integration to submission quality
Feb 2018 — Jun 2020
Boehringer Ingelheim · Ridgefield, CT
Statistical Programmer
- ISE/ISS pooled datasets for oncology and cardiovascular regulatory submissions
- Reusable SAS macro libraries; CDISC TAUG development; DaVinci open-source R Shiny visualization; automated Pinnacle 21 compliance workflows
May 2017 — Feb 2018
Talentech · Monmouth Junction, NJ
SAS Programmer
- TLF generation on standard ADaM datasets (ADSL, ADLB, ADAE, ADTTE, ADRS) with strict CDISC compliance; core SAS macro development
Selected Highlights
Regulatory
First-in-class NDA, end to end
Statistical programming lead from Phase 2b through pivotal Phase 3 to FDA approval of LIPFENDRA® — Define-XML v2.1, reviewer's guides, eCTD, and ADRG delivered with full CDISC compliance.
GenAI
Enterprise AI automation
Led GenAI automation initiatives including an enterprise LLM/RAG platform; architected agentic systems (ClinAgent, GxP-Agent) published in peer-reviewed journals.
Open Source
Tools the industry uses
Creator of py4csr (PyPI, Zenodo DOI); the public eSubmission-Benchmark dataset, cited at PharmaSUG 2026.
Visualization
Enterprise analytics platforms
Enterprise data visualization and AI tools for automated clinical trial reporting; interactive DMC reporting on the metalite/forestly/boxly R ecosystem.
Expertise
Technical Capabilities
End-to-end clinical trial programming — from CDISC standards and regulatory submissions to GenAI automation and validated analytics platforms.
Statistical Programming
- CDISC standards — SDTM, ADaM, Define-XML v2.1
- TLF generation & validation
- Pinnacle 21 compliance workflows
- Group sequential design reporting
Languages & Packages
- SAS — Base & Advanced certified
- R — admiral, Shiny, metalite suite
- Python — py4csr, Dash, pandas
- SQL, Neo4j / Cypher
AI / LLM
- Agentic systems — ClinAgent, GxP-Agent
- Enterprise LLM / RAG platforms
- Small-LLM fine-tuning for code generation
- Knowledge-graph-constrained validation
Regulatory & eSubmission
- NDA/BLA eCTD packages, ADRG, reviewer's guides
- 21 CFR Part 11 validated environments
- GAMP 5 risk-based validation
- Open-source tooling (py4csr, eSubmission-Benchmark)
Visualization & Platforms
- Interactive DMC reporting (metalite, forestly, boxly)
- Python Dash & R Shiny applications
- Enterprise analytics platform architecture
- Docker, CI/CD, Git
Leadership
- Program-level statistical programming lead
- NDA submission lead (FDA-approved)
- CRO programming-vendor oversight
- Mentoring & R/Python training
Contact
Get in Touch
Open to conversations about statistical programming, AI/LLM automation in clinical trials, and research collaboration.
Areas of interest: Statistical programming · AI/LLM research · Clinical trials · Data science · Analytics leadership