TLF Studio
— shell → R → RTF → PDF, end-to-end
checking…
TLF request
📚
Try a sample TLF shell
· 13 mock shells, no PHI · click any to load
▶
demographics
disposition
AE summary
📑 Upload ARS JSON
CDISC Analysis Results Standard (siera / atlas)
Shell text / description
📎 Upload PDF/RTF/DOCX/image
or
📋 or paste a screenshot anywhere (Ctrl+V) · or drag & drop
Table type
demographics
disposition
ae_summary
lab_summary
exposure
other
Table label
Datasets (comma-sep)
Datasets are auto-staged; v2 dummy data is enriched with derived CDISC vars (EOSSTT, DCSREAS, AESOC).
Extra notes (optional)
Optional: rebind to your data
(after generating, paste your CSV header to swap canonical names → your column names)
Rebind R to my variables
🔧 Build shell_desc by form (no JSON, no vision needed)
Pick a layout, set a title, add rows. Each row maps to one statistical block. The composer runs immediately — no LLM, no RAG.
Layout:
pivot_wide_by_treatment (rows × treatment cols)
pivot_wide_by_visit (visit rows × treatment cols)
nested_soc_pt (SOC ▸ PT × treatment)
shift_2d_layout (baseline grade × post grade)
figure_layout (for plots)
Label:
Title:
Population:
safety (saffl == "Y")
itt (ittfl == "Y")
fas (fasfl == "Y")
per-protocol (ppfl == "Y")
Datasets:
Row groups
+ Add row
Use this shell_desc
Preview JSON
RAG mode
· retrieve a similar production SAS program (745-corpus) → adapt to R via GLM-5.1 → verify+repair if needed
Generate
Generated R code
Download .R
Download .SAS
QC findings
hide
Preview
Download .RTF
Download .PDF
The PDF preview will appear here
after you generate a TLF.
stderr
stdout
close ✕
📖 Guide