Replicate ECG Results at Planned Time Points

1library(yamaa)23yamaa_spec("spec.yaml")$run()$output

1import yamaa23yamaa.spec("spec.yaml").run().output

1# Reference solution for the yamaa benchmark sdtm-eg-ecg-findings (R track).2#3# Written from the benchmark's full prompt and its inputs alone, as an agent4# would write it. Harbor's oracle agent runs it to check that the prompt can5# be solved and that the grader scores a correct answer 1.67library(dplyr, warn.conflicts = FALSE)8library(readr)9library(tidyr)1011odm <- read_csv(12  "/app/input/odm.csv",13  col_types = cols(.default = col_character()),14  na = ""15)1617# One replicate ECG is one repeat of the ECG item group on a time point's18# form; its items become columns.19ecgs <- odm |>20  filter(FormOID %in% c("FO.EG_PREDOSE", "FO.EG_2H")) |>21  select(StudyOID, SubjectKey, StudyEventOID, FormOID, ItemGroupRepeatKey,22         ItemOID, Value) |>23  pivot_wider(names_from = ItemOID, values_from = Value)2425# One record per test; a blank result gives no record.26qt <- ecgs |>27  transmute(StudyOID, SubjectKey, StudyEventOID, FormOID, ItemGroupRepeatKey,28            EGDTC = IT.EG.EGDTC,29            EGTESTCD = "QTAG", EGTEST = "QT Interval, Aggregate",30            EGORRES = IT.EG.QT, EGORRESU = "msec")31interpretation <- ecgs |>32  transmute(StudyOID, SubjectKey, StudyEventOID, FormOID, ItemGroupRepeatKey,33            EGDTC = IT.EG.EGDTC,34            EGTESTCD = "INTP", EGTEST = "Interpretation",35            EGORRES = IT.EG.INTP, EGORRESU = NA_character_)3637eg <- bind_rows(qt, interpretation) |>38  filter(!is.na(EGORRES), trimws(EGORRES) != "") |>39  mutate(40    DOMAIN = "EG",41    STUDYID = StudyOID,42    USUBJID = SubjectKey,43    EGSTRESC = EGORRES,44    EGSTRESN = if_else(EGTESTCD == "QTAG", suppressWarnings(as.numeric(EGORRES)),45                       NA_real_),46    EGSTRESU = EGORRESU,47    EGREPNUM = as.integer(ItemGroupRepeatKey),48    VISIT = case_when(StudyEventOID == "SE.DAY1" ~ "DAY 1"),49    EGTPT = case_when(50      FormOID == "FO.EG_PREDOSE" ~ "PRE-DOSE",51      FormOID == "FO.EG_2H" ~ "2 H POST-DOSE"52    ),53    EGTPTNUM = case_when(54      FormOID == "FO.EG_PREDOSE" ~ 1L,55      FormOID == "FO.EG_2H" ~ 2L56    )57  ) |>58  arrange(STUDYID, USUBJID, EGTPTNUM, EGREPNUM, EGTESTCD) |>59  group_by(STUDYID, USUBJID) |>60  mutate(EGSEQ = row_number()) |>61  ungroup() |>62  select(DOMAIN, STUDYID, USUBJID, EGSEQ, EGTESTCD, EGTEST, EGORRES, EGORRESU,63         EGSTRESC, EGSTRESN, EGSTRESU, EGREPNUM, VISIT, EGDTC, EGTPT,64         EGTPTNUM)6566dir.create("/app/output", showWarnings = FALSE)67write_csv(eg, "/app/output/eg.csv", na = "")

1# Reference solution for the yamaa benchmark sdtm-eg-ecg-findings (Python2# track).3#4# Written from the benchmark's full prompt and its inputs alone, as an agent5# would write it. Harbor's oracle agent runs it to check that the prompt can6# be solved and that the grader scores a correct answer 1.78import csv910ODM_CSV = "/app/input/odm.csv"11OUTPUT_CSV = "/app/output/eg.csv"1213VISITS = {"SE.DAY1": "DAY 1"}14TIME_POINTS = {15    "FO.EG_PREDOSE": ("PRE-DOSE", 1),16    "FO.EG_2H": ("2 H POST-DOSE", 2),17}18# Item OID: test code, test name, unit (None for a character result).19TESTS = {20    "IT.EG.QT": ("QTAG", "QT Interval, Aggregate", "msec"),21    "IT.EG.INTP": ("INTP", "Interpretation", None),22}23DATETIME_ITEM = "IT.EG.EGDTC"2425COLUMNS = [26    "DOMAIN", "STUDYID", "USUBJID", "EGSEQ", "EGTESTCD", "EGTEST", "EGORRES",27    "EGORRESU", "EGSTRESC", "EGSTRESN", "EGSTRESU", "EGREPNUM", "VISIT",28    "EGDTC", "EGTPT", "EGTPTNUM",29]303132def main():33    with open(ODM_CSV, newline="") as f:34        odm = list(csv.DictReader(f))3536    # One replicate ECG is one repeat of the ECG item group on a time point's37    # form; collect its items.38    ecgs = {}39    for row in odm:40        if row["FormOID"] not in TIME_POINTS:41            continue42        key = (43            row["StudyOID"],44            row["SubjectKey"],45            row["StudyEventOID"],46            row["FormOID"],47            int(row["ItemGroupRepeatKey"]),48        )49        ecgs.setdefault(key, {})[row["ItemOID"]] = row["Value"].strip()5051    records = []52    for (study, subject, event, form, repeat), items in ecgs.items():53        tpt, tptnum = TIME_POINTS[form]54        for item, (testcd, test, unit) in TESTS.items():55            result = items.get(item, "")56            if result == "":57                continue  # a blank result gives no record58            records.append(59                {60                    "DOMAIN": "EG",61                    "STUDYID": study,62                    "USUBJID": subject,63                    "EGTESTCD": testcd,64                    "EGTEST": test,65                    "EGORRES": result,66                    "EGORRESU": unit or "",67                    "EGSTRESC": result,68                    "EGSTRESN": result if unit else "",69                    "EGSTRESU": unit or "",70                    "EGREPNUM": repeat,71                    "VISIT": VISITS[event],72                    "EGDTC": items.get(DATETIME_ITEM, ""),73                    "EGTPT": tpt,74                    "EGTPTNUM": tptnum,75                }76            )7778    records.sort(79        key=lambda r: (80            r["STUDYID"], r["USUBJID"], r["EGTPTNUM"], r["EGREPNUM"],81            r["EGTESTCD"],82        )83    )84    sequence = {}85    for record in records:86        subject = (record["STUDYID"], record["USUBJID"])87        sequence[subject] = sequence.get(subject, 0) + 188        record["EGSEQ"] = sequence[subject]8990    with open(OUTPUT_CSV, "w", newline="") as f:91        writer = csv.DictWriter(f, fieldnames=COLUMNS)92        writer.writeheader()93        writer.writerows(records)949596if __name__ == "__main__":97    main()

1/* opensas reference solution for the yamaa benchmark sdtm-eg-ecg-findings.2   Reads the benchmark inputs and derives the requested output.3   Execute with the existing opensas runtime: opensas /app/output/result.sas. */45data odm;6  length StudyOID MetaDataVersionOID SubjectKey StudyEventOID StudyEventRepeatKey FormOID FormRepeatKey ItemGroupOID ItemGroupRepeatKey ItemOID Value $1024;7  infile "/app/input/odm.csv" dsd dlm="," firstobs=2 truncover lrecl=32767;8  input StudyOID : $char1024. MetaDataVersionOID : $char1024. SubjectKey : $char1024. StudyEventOID : $char1024. StudyEventRepeatKey : $char1024. FormOID : $char1024. FormRepeatKey : $char1024. ItemGroupOID : $char1024. ItemGroupRepeatKey : $char1024. ItemOID : $char1024. Value : $char1024.;9run;1011/* One replicate ECG is one repeat of the ECG item group on a time point's12   form: collect its date and time, QT interval and interpretation. */13proc sort data=odm out=ordered; by StudyOID SubjectKey StudyEventOID FormOID ItemGroupRepeatKey; run;14data wide;15  length EGDTC QT INTP $1024;16  retain EGDTC QT INTP;17  set ordered;18  where FormOID in ('FO.EG_PREDOSE','FO.EG_2H');19  by StudyOID SubjectKey StudyEventOID FormOID ItemGroupRepeatKey;20  if first.ItemGroupRepeatKey then do;21    EGDTC=''; QT=''; INTP='';22  end;23  if ItemOID='IT.EG.EGDTC' then EGDTC=Value;24  if ItemOID='IT.EG.QT' then QT=Value;25  if ItemOID='IT.EG.INTP' then INTP=Value;26  if last.ItemGroupRepeatKey then output;27run;2829/* One record per test; a blank result gives no record. */30data prepared; set wide;31  length DOMAIN $2 STUDYID USUBJID EGTESTCD EGTEST EGORRES EGORRESU EGSTRESC EGSTRESU VISIT EGTPT $1024;32  DOMAIN='EG'; STUDYID=StudyOID; USUBJID=SubjectKey;33  VISIT=''; if StudyEventOID='SE.DAY1' then VISIT='DAY 1';34  EGTPT=''; EGTPTNUM=.;35  if FormOID='FO.EG_PREDOSE' then do; EGTPT='PRE-DOSE'; EGTPTNUM=1; end;36  if FormOID='FO.EG_2H' then do; EGTPT='2 H POST-DOSE'; EGTPTNUM=2; end;37  EGREPNUM=input(ItemGroupRepeatKey,best32.);38  if not missing(QT) then do;39    EGTESTCD='QTAG'; EGTEST='QT Interval, Aggregate';40    EGORRES=strip(QT); EGORRESU='msec'; EGSTRESC=EGORRES;41    EGSTRESN=input(EGORRES,best32.); EGSTRESU='msec';42    output;43  end;44  if not missing(INTP) then do;45    EGTESTCD='INTP'; EGTEST='Interpretation';46    EGORRES=strip(INTP); EGORRESU=''; EGSTRESC=EGORRES;47    EGSTRESN=.; EGSTRESU='';48    output;49  end;50run;51proc sort data=prepared; by STUDYID USUBJID EGTPTNUM EGREPNUM EGTESTCD; run;52data result; set prepared; by STUDYID USUBJID; retain EGSEQ;53  if first.USUBJID then EGSEQ=0;54  EGSEQ+1;55run;5657proc sql;58  create table final as select DOMAIN, STUDYID, USUBJID, EGSEQ, EGTESTCD, EGTEST, EGORRES, EGORRESU, EGSTRESC, EGSTRESN, EGSTRESU, EGREPNUM, VISIT, EGDTC, EGTPT, EGTPTNUM from result;59quit;60proc export data=final outfile="/app/output/eg.csv" dbms=csv replace; run;