yamaa 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 meansREADME.mdGoal: group each adverse event by safety topic, showing which coded term falls under two Standardized MedDRA Query (SMQ) groupings and one sponsor-defined customized query (CQ) grouping.
Input: one row per collected adverse event with the reported term
AETERM and the Medical Dictionary for Regulatory Activities
(MedDRA) coded term AEDECOD. A query dictionary lists each coded
term under its query groupings.
Variables:
Each query grouping has its own fixed set of columns.
SMQ01NAM: name of the first standardized query grouping the coded term belongs to; empty when the term is not in it.SMQ01CD: dictionary code of the first standardized query grouping; empty when the term is not in it.SMQ01SC: whether membership is broad or narrow scope (BROAD/NARROW); empty when the term is not in it.SMQ02NAM: name of the second standardized query grouping the coded term belongs to; empty when the term is not in it.SMQ02CD: dictionary code of the second standardized query grouping; empty when the term is not in it.SMQ02SC: whether membership is broad or narrow scope (BROAD/NARROW); empty when the term is not in it.CQ01NAM: name of the sponsor-defined grouping the term belongs to; empty when not listed. No code or scope applies.
Note: an event can sit in a standardized grouping and in the
sponsor grouping at the same time, each shown in its own place:
Stevens-Johnson syndrome carries both the skin-reaction grouping and
the sponsor grouping. Two events with the same coded term always show
the same grouping entries: both nausea events carry the sponsor
grouping. An event still awaiting coding, with no AEDECOD, belongs
to no grouping.
Inputs
Under review2 source filesinput/ae.csv
Edit10 rows| STUDYID | USUBJID | AESEQ | AETERM | AEDECOD |
|---|---|---|---|---|
| PILOT7 | 1 | Rash all over | RASH | |
| PILOT7 | 2 | Feeling sick | NAUSEA | |
| PILOT7 | 1 | Jaundice and dark urine | HEPATITIS | |
| PILOT7 | 2 | Loose stools | DIARRHOEA | |
| PILOT7 | 1 | Headache | HEADACHE | |
| PILOT7 | 2 | Odd sensation in the left arm | — | |
| PILOT7 | 1 | Severe skin blistering | STEVENS-JOHNSON SYNDROME | |
| PILOT7 | 1 | Recurrent nausea | NAUSEA | |
| PILOT7 | 2 | Acute liver failure | HEPATIC FAILURE | |
| PILOT7 | 3 | Target lesions on trunk | ERYTHEMA MULTIFORME |
input/queries.csv
Edit8 rows| PREFIX | GRPNAME | GRPID | SCOPE | TERM |
|---|---|---|---|---|
| SMQ01 | Severe Cutaneous Adverse Reactions | 20000020 | BROAD | RASH |
| SMQ01 | Severe Cutaneous Adverse Reactions | 20000020 | BROAD | ERYTHEMA MULTIFORME |
| SMQ01 | Severe Cutaneous Adverse Reactions | 20000020 | NARROW | STEVENS-JOHNSON SYNDROME |
| SMQ02 | Hepatic Disorders | 20000006 | NARROW | HEPATITIS |
| SMQ02 | Hepatic Disorders | 20000006 | NARROW | HEPATIC FAILURE |
| CQ01 | Sponsor Adverse Events of Special Interest | — | — | DIARRHOEA |
| CQ01 | Sponsor Adverse Events of Special Interest | — | — | NAUSEA |
| CQ01 | Sponsor Adverse Events of Special Interest | — | — | STEVENS-JOHNSON SYNDROME |
Expected output
Under review10 expected rowsexpected/adae.csv
Edit10 rows| STUDYID | USUBJID | AESEQ | AETERM | AEDECOD | SMQ01NAM | SMQ01CD | SMQ01SC | SMQ02NAM | SMQ02CD | SMQ02SC | CQ01NAM |
|---|---|---|---|---|---|---|---|---|---|---|---|
| PILOT7 | 1 | Rash all over | RASH | Severe Cutaneous Adverse Reactions | 20000020 | BROAD | — | — | — | — | |
| PILOT7 | 2 | Feeling sick | NAUSEA | — | — | — | — | — | — | Sponsor Adverse Events of Special Interest | |
| PILOT7 | 1 | Jaundice and dark urine | HEPATITIS | — | — | — | Hepatic Disorders | 20000006 | NARROW | — | |
| PILOT7 | 2 | Loose stools | DIARRHOEA | — | — | — | — | — | — | Sponsor Adverse Events of Special Interest | |
| PILOT7 | 1 | Headache | HEADACHE | — | — | — | — | — | — | — | |
| PILOT7 | 2 | Odd sensation in the left arm | — | — | — | — | — | — | — | — | |
| PILOT7 | 1 | Severe skin blistering | STEVENS-JOHNSON SYNDROME | Severe Cutaneous Adverse Reactions | 20000020 | NARROW | — | — | — | Sponsor Adverse Events of Special Interest | |
| PILOT7 | 1 | Recurrent nausea | NAUSEA | — | — | — | — | — | — | Sponsor Adverse Events of Special Interest | |
| PILOT7 | 2 | Acute liver failure | HEPATIC FAILURE | — | — | — | Hepatic Disorders | 20000006 | NARROW | — | |
| PILOT7 | 3 | Target lesions on trunk | ERYTHEMA MULTIFORME | Severe Cutaneous Adverse Reactions | 20000020 | BROAD | — | — | — | — |
Comments and assessment materials
| Dataset Name | Variable Name | Variable Label | Type | Codelist/Controlled Terms | Core | Computational Method | Origin |
|---|---|---|---|---|---|---|---|
| ADAE | STUDYID | Study Identifier | character | Copy value from AE.STUDYID. | Predecessor | ||
| ADAE | USUBJID | Unique Subject Identifier | character | Copy value from AE.USUBJID. | Predecessor | ||
| ADAE | AESEQ | Sequence Number | numeric | Copy value from AE.AESEQ. | Predecessor | ||
| ADAE | AETERM | Reported Term for the Adverse Event | character | Copy value from AE.AETERM. | Predecessor | ||
| ADAE | AEDECOD | Dictionary-Derived Term | character | Copy value from AE.AEDECOD. | Predecessor | ||
| ADAE | SMQ01NAM | Standardized MedDRA Query 01 Name | character | Take GRPNAME from the QUERY record whose TERM equals AEDECOD, where QUERY.PREFIX = 'SMQ01', and no match yields blank. | Derived | ||
| ADAE | SMQ01CD | Standardized MedDRA Query 01 Code | numeric | Take GRPID from the QUERY record whose TERM equals AEDECOD, where QUERY.PREFIX = 'SMQ01', and no match yields blank. | Derived | ||
| ADAE | SMQ01SC | Standardized MedDRA Query 01 Scope | character | BROAD, NARROW | Take SCOPE from the QUERY record whose TERM equals AEDECOD, where QUERY.PREFIX = 'SMQ01', and no match yields blank. | Derived | |
| ADAE | SMQ02NAM | Standardized MedDRA Query 02 Name | character | Take GRPNAME from the QUERY record whose TERM equals AEDECOD, where QUERY.PREFIX = 'SMQ02', and no match yields blank. | Derived | ||
| ADAE | SMQ02CD | Standardized MedDRA Query 02 Code | numeric | Take GRPID from the QUERY record whose TERM equals AEDECOD, where QUERY.PREFIX = 'SMQ02', and no match yields blank. | Derived | ||
| ADAE | SMQ02SC | Standardized MedDRA Query 02 Scope | character | BROAD, NARROW | Take SCOPE from the QUERY record whose TERM equals AEDECOD, where QUERY.PREFIX = 'SMQ02', and no match yields blank. | Derived | |
| ADAE | CQ01NAM | Customized Query 01 Name | character | Take GRPNAME from the QUERY record whose TERM equals AEDECOD, where QUERY.PREFIX = 'CQ01', and no match yields blank. | Derived |
| Variable | Permitted values |
|---|---|
| SMQ01SC | BROAD, NARROW |
| SMQ02SC | BROAD, NARROW |
| Version | Date | Author | Description | Reviewer | Sign-off |
|---|---|---|---|---|---|
| 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 |
prompts/adam-adae-query-flags/full.md
Edit18 linesFollowing CDISC ADaM standards, use the provided AE and QUERY datasets to create an ADAE dataset with one record per adverse event.
The output dataset should contain the following columns in this order: STUDYID, USUBJID, AESEQ, AETERM, AEDECOD, SMQ01NAM, SMQ01CD, SMQ01SC, SMQ02NAM, SMQ02CD, SMQ02SC, CQ01NAM
Show each coded term's entries from the query dictionary in their own slots. Each standardized slot shows its grouping name, dictionary code, and scope together: BROAD or NARROW. The sponsor slot shows the grouping name only. An event can sit in a standardized grouping and in the sponsor grouping at the same time, each shown in its own place. Two events with the same coded term always show the same grouping entries. An event still awaiting coding, with no AEDECOD, belongs to no grouping, so the grouping columns have no value.
Read the source datasets from /app/input and save the completed dataset as /app/output/adae.csv.
prompts/adam-adae-query-flags/conventions.md
Edit11 linesFollowing CDISC ADaM standards, use the provided AE and QUERY datasets to create an ADAE dataset with one record per adverse event.
The output dataset should contain the following columns in this order: STUDYID, USUBJID, AESEQ, AETERM, AEDECOD, SMQ01NAM, SMQ01CD, SMQ01SC, SMQ02NAM, SMQ02CD, SMQ02SC, CQ01NAM
The scope of each standardized slot is BROAD or NARROW.
Read the source datasets from /app/input and save the completed dataset as /app/output/adae.csv.
prompts/adam-adae-query-flags/brief.md
Edit9 linesFollowing CDISC ADaM standards, use the provided AE and QUERY datasets to create an ADAE dataset with one record per adverse event.
The output dataset should contain the following columns in this order: STUDYID, USUBJID, AESEQ, AETERM, AEDECOD, SMQ01NAM, SMQ01CD, SMQ01SC, SMQ02NAM, SMQ02CD, SMQ02SC, CQ01NAM
Read the source datasets from /app/input and save the completed dataset as /app/output/adae.csv.
run.R
Edit4 lineslibrary(yamaa)adae <- yamaa_domain("spec.yaml")$outputadaerun.py
Edit4 linesimport yamaaadae = yamaa.yamaa_domain("spec.yaml").outputadaesolutions/adam-adae-query-flags/result.R
Edit44 lines# Reference solution for the yamaa benchmark adam-adae-query-flags (R track).## Written from the benchmark's full prompt and its inputs alone, as an agent# would write it. Harbor's oracle agent runs it to check that the prompt can# be solved and that the grader scores a correct answer 1.library(dplyr, warn.conflicts = FALSE)library(readr)ae <- read_csv( "/app/input/ae.csv", col_types = cols(AESEQ = col_integer(), .default = col_character()))queries <- read_csv( "/app/input/queries.csv", col_types = cols(GRPID = col_integer(), .default = col_character()))# Each standardized slot shows its grouping name, dictionary code, and# scope together; the sponsor slot shows the grouping name only. Two# events with the same coded term always show the same entries.smq01 <- queries |> filter(PREFIX %in% "SMQ01") |> select(TERM, SMQ01NAM = GRPNAME, SMQ01CD = GRPID, SMQ01SC = SCOPE)smq02 <- queries |> filter(PREFIX %in% "SMQ02") |> select(TERM, SMQ02NAM = GRPNAME, SMQ02CD = GRPID, SMQ02SC = SCOPE)cq01 <- queries |> filter(PREFIX %in% "CQ01") |> select(TERM, CQ01NAM = GRPNAME)adae <- ae |> left_join(smq01, by = c("AEDECOD" = "TERM")) |> left_join(smq02, by = c("AEDECOD" = "TERM")) |> left_join(cq01, by = c("AEDECOD" = "TERM")) |> select( STUDYID, USUBJID, AESEQ, AETERM, AEDECOD, SMQ01NAM, SMQ01CD, SMQ01SC, SMQ02NAM, SMQ02CD, SMQ02SC, CQ01NAM )dir.create("/app/output", showWarnings = FALSE)write_csv(adae, "/app/output/adae.csv", na = "")solutions/adam-adae-query-flags/result.py
Edit50 lines# Reference solution for the yamaa benchmark adam-adae-query-flags (Python track).## Written from the benchmark's full prompt and its inputs alone, as an agent# would write it. Harbor's oracle agent runs it to check that the prompt can# be solved and that the grader scores a correct answer 1.from pathlib import Pathimport polars as plae = pl.read_csv("/app/input/ae.csv", infer_schema=False).with_columns( pl.col("AESEQ").cast(pl.Int64))queries = pl.read_csv("/app/input/queries.csv", infer_schema=False).with_columns( pl.col("GRPID").cast(pl.Int64))# Each standardized slot shows its grouping name, dictionary code, and# scope together; the sponsor slot shows the grouping name only. Two# events with the same coded term always show the same entries.smq01 = queries.filter(pl.col("PREFIX") == "SMQ01").select( "TERM", SMQ01NAM="GRPNAME", SMQ01CD="GRPID", SMQ01SC="SCOPE")smq02 = queries.filter(pl.col("PREFIX") == "SMQ02").select( "TERM", SMQ02NAM="GRPNAME", SMQ02CD="GRPID", SMQ02SC="SCOPE")cq01 = queries.filter(pl.col("PREFIX") == "CQ01").select("TERM", CQ01NAM="GRPNAME")adae = ( ae.join(smq01, left_on="AEDECOD", right_on="TERM", how="left", maintain_order="left") .join(smq02, left_on="AEDECOD", right_on="TERM", how="left", maintain_order="left") .join(cq01, left_on="AEDECOD", right_on="TERM", how="left", maintain_order="left") .select( "STUDYID", "USUBJID", "AESEQ", "AETERM", "AEDECOD", "SMQ01NAM", "SMQ01CD", "SMQ01SC", "SMQ02NAM", "SMQ02CD", "SMQ02SC", "CQ01NAM", ))Path("/app/output").mkdir(exist_ok=True)adae.write_csv("/app/output/adae.csv")solutions/adam-adae-query-flags/result.sas
Edit30 lines/* opensas reference solution for the yamaa benchmark adam-adae-query-flags. Reads the benchmark inputs and derives the requested output. Execute with the existing opensas runtime: opensas /app/output/result.sas. */data ae; length STUDYID USUBJID AETERM AEDECOD $1024; infile "/app/input/ae.csv" dsd dlm="," firstobs=2 truncover lrecl=32767; input STUDYID : $char1024. USUBJID : $char1024. AESEQ AETERM : $char1024. AEDECOD : $char1024.;run;data queries; length PREFIX GRPNAME GRPID SCOPE TERM $1024; infile "/app/input/queries.csv" dsd dlm="," firstobs=2 truncover lrecl=32767; input PREFIX : $char1024. GRPNAME : $char1024. GRPID : $char1024. SCOPE : $char1024. TERM : $char1024.;run;proc sql; create table result as select a.*, b.GRPNAME as SMQ01NAM,b.GRPID as SMQ01CD,b.SCOPE as SMQ01SC, c.GRPNAME as SMQ02NAM,c.GRPID as SMQ02CD,c.SCOPE as SMQ02SC, d.GRPNAME as CQ01NAM from ae as a left join queries as b on a.AEDECOD=b.TERM and b.PREFIX='SMQ01' and a.AEDECOD ne '' left join queries as c on a.AEDECOD=c.TERM and c.PREFIX='SMQ02' and a.AEDECOD ne '' left join queries as d on a.AEDECOD=d.TERM and d.PREFIX='CQ01' and a.AEDECOD ne '';quit;proc sql; create table final as select STUDYID, USUBJID, AESEQ, AETERM, AEDECOD, SMQ01NAM, SMQ01CD, SMQ01SC, SMQ02NAM, SMQ02CD, SMQ02SC, CQ01NAM from result;quit;proc export data=final outfile="/app/output/adae.csv" dbms=csv replace; run;
Comments
Signed in with GitHub; saved in repository Discussions