Execution Lifecycle

Review ask

As a clinical trial statistician or programmer, challenge anything in this benchmark: describe a problem in Comments, or propose a fix with the Edit button on any source file.

Task: Would a real study need this derivation, specified this way? Does it follow CDISC ADaM: variable names, controlled terms, record structure? Flag any rule that is ambiguous or that a sponsor would handle differently.
Inputs: Do they look like data collected in a real trial, and do they include the situations you would expect to meet in one?
Expected output: Would independent QC programming reproduce every value from these inputs and rules?

The other tabs beside Comments are background for assessment, not under review.

Summary

EditUnder reviewWhat this benchmark means

README.md

Goal: show the order a specification runs in: rows are built first, then columns are added on top of the finished rows, and rows discarded along the way never reach the final checks.

Input: one spec.yaml over one lab-results file. Each record carries a subject (USUBJID), a visit (VISIT), a test code (ALT or AST), and a result that arrives as text.

Behavior:

A record with no result never becomes a row: it is dropped before any row is built. Each surviving record becomes one component row carrying its test code, the test's full name (PARAM), and its result (AVAL).
One summary row per subject-visit carries that visit's lowest result. The visit whose results are all missing is discarded after its summary is computed, so the final completeness check on AVAL passes: it never sees the discarded row.
Text results become numbers while the rows are built, so the two columns added afterwards read numbers, never text: DOUBLE holds twice the result, and FLAG marks results above 50 with H.
Every row carries the study identifier YAMAA-01 in STUDYID.

Note: building rows may change the row count while adding columns may not: each finished row gets exactly one DOUBLE and one FLAG.

YAML specification

1 spec file

spec.yaml

Edit99 lines

1schema_version: "1.0"2domain: ADLB3keys: [USUBJID, VISIT, PARAMCD]4input:5  # Results arrive as text: the row-building stage converts each6  # derived value to its declared type while the rows are built.7  LB: {path: input/lb.csv, types: {LBSTRESN: str}}8base: LB910output:11  path: adlb.csv12  columns: [STUDYID, USUBJID, VISIT, PARAMCD, PARAM, AVAL, DOUBLE, FLAG]1314columns:15  # Uniform derivations shared by both row templates live at column16  # level and run in declaration order.17  - name: STUDYID18    type: str19    label: Study Identifier20    derivation: {literal: YAMAA-01}2122  - name: USUBJID23    type: str24    label: Unique Subject Identifier25    derivation: LB.USUBJID2627  - name: VISIT28    type: int29    label: Visit Number30    derivation: LB.VISIT3132  - name: PARAMCD33    type: str34    label: Parameter Code3536  - name: PARAM37    type: str38    label: Parameter3940  # A row-stage column: each template derives it while building its41  # rows, converting the text result to the declared float there. Its42  # stage-4 check runs over the completed column, which never contains43  # the discarded visit below.44  - name: AVAL45    type: float46    label: Analysis Value47    verifications:48      - not_missing: {}4950  # Column stage: enriches the built rows with exactly one value each51  # and may not change the row count. The arithmetic reads AVAL's52  # converted float value; a text value would fail the53  # no-implicit-conversion rule.54  - name: DOUBLE55    type: float56    label: Twice the Analysis Value57    derivation:58      compute:59        expr: "2 * AVAL"6061  - name: FLAG62    type: str63    label: High Value Flag64    derivation:65      case:66        - when: "AVAL > 50"67          then: {literal: H}68        - otherwise: {literal: N}6970rows:71  # The record screen runs before any row value is built: records with72  # no result never become rows.73  - id: components74    filter: "LB.LBSTRESN IS NOT NULL"75    derivations:76      PARAMCD: LB.LBTESTCD77      PARAM:78        case:79          - when: "LB.LBTESTCD = 'ALT'"80            then: {literal: Alanine Aminotransferase}81          - otherwise: {literal: Aspartate Aminotransferase}82      AVAL: LB.LBSTRESN8384  # One candidate row per subject-visit. The visit screen runs after85  # the lowest result is reduced: the visit whose results are all86  # missing is discarded, so the final checks below never see it.87  - id: visit_low88    group_by: [LB.USUBJID, LB.VISIT]89    filter: "AVAL IS NOT NULL"90    derivations:91      PARAMCD: {literal: LOW}92      PARAM: {literal: Lowest Result of Visit}93      AVAL:94        aggregate:95          expr: "MIN(LB.LBSTRESN)"9697verifications:98  - unique:99      columns: [USUBJID, VISIT, PARAMCD]

Inputs

Under review1 source file

input/lb.csv

Edit8 rows

input/lb.csv
VISIT	LBTESTCD	LBSTRESN
1	ALT	32
1	AST	45
2	ALT	28
2	AST	—
3	ALT	—
3	AST	—
1	ALT	60
1	AST	70

Expected output

Under review8 expected rows

expected/adlb.csv

Edit8 rows

expected/adlb.csv
STUDYID	VISIT	PARAMCD	PARAM	AVAL	DOUBLE	FLAG
YAMAA-01	1	ALT	Alanine Aminotransferase	32	64	N
YAMAA-01	1	AST	Aspartate Aminotransferase	45	90	N
YAMAA-01	2	ALT	Alanine Aminotransferase	28	56	N
YAMAA-01	1	ALT	Alanine Aminotransferase	60	120	H
YAMAA-01	1	AST	Aspartate Aminotransferase	70	140	H
YAMAA-01	1	LOW	Lowest Result of Visit	32	64	N
YAMAA-01	2	LOW	Lowest Result of Visit	28	56	N
YAMAA-01	1	LOW	Lowest Result of Visit	60	120	H

Shaded columns are derived— = missing value

Comments and assessment materials

Dataset Name	Variable Name	Variable Label	Type	Computational Method	Origin
ADLB	STUDYID	Study Identifier	character	Set constant value "YAMAA-01".	Assigned
ADLB	USUBJID	Unique Subject Identifier	character	Copy value from LB.USUBJID.	Predecessor
ADLB	VISIT	Visit Number	numeric	Copy value from LB.VISIT.	Predecessor
ADLB	PARAMCD	Parameter Code	character	Set per constructed row; see the Row construction sheet.	Derived
ADLB	PARAM	Parameter	character	Set per constructed row; see the Row construction sheet.	Derived
ADLB	AVAL	Analysis Value	numeric	Set per constructed row; see the Row construction sheet.	Derived
ADLB	DOUBLE	Twice the Analysis Value	numeric	Compute 2 * AVAL.	Derived
ADLB	FLAG	High Value Flag	character	If AVAL > 50 then "H"; otherwise "N".	Derived

Template	Include when	PARAMCD	PARAM	AVAL
components	LB.LBSTRESN IS NOT NULL	Copy LB.LBTESTCD	If LB.LBTESTCD = 'ALT' then "Alanine Aminotransferase"; otherwise "Aspartate Aminotransferase"	Copy LB.LBSTRESN
visit_low	AVAL IS NOT NULL	Constant "LOW"	Constant "Lowest Result of Visit"	Aggregate MIN(LB.LBSTRESN) over records matching on the shared output keys

Version	Date	Author	Description
1.3	2026-09-29	yamaa docs build	Show study codelist names, external dictionaries, and ISO 8601 formats in variable rows; list the document's codelist values and metadata
1.2	2026-09-29	yamaa docs build	Classify ADaM copies from a declared input as Predecessor and recodes as Derived, and SDTM recodes of locally derived columns as Derived
1.1	2026-09-29	yamaa docs build	Resolve lookups into the donor record they select, and give every derivation shape a plain-language rule instead of its raw YAML
1.0		yamaa docs build	Initial generation from spec.yaml

run.R

Edit4 lines

1library(yamaa)23adlb <- yamaa_domain("spec.yaml")$output4adlb

run.py

Edit4 lines

1import yamaa23adlb = yamaa.yamaa_domain("spec.yaml").output4adlb

input/lb.csv

expected/adlb.csv

Comments

Mapping spec

yamaa code

run.R

run.py