Build One Record per Collected Lab Result

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

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

1# Reference solution for the yamaa benchmark sdtm-lb-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)1516key_cols <- c(17  "StudyOID", "SubjectKey", "StudyEventOID", "StudyEventRepeatKey",18  "ItemGroupOID", "ItemGroupRepeatKey"19)2021dates <- odm |>22  filter(ItemOID == "IT.LB.LBDTC") |>23  select(all_of(key_cols), LBDTC = Value)2425lb <- odm |>26  filter(ItemOID %in% c("IT.LB.CALCIUM", "IT.LB.CREAT")) |>27  left_join(dates, by = key_cols) |>28  mutate(Value = na_if(Value, "")) |>29  filter(!is.na(Value)) |>30  mutate(31    LBTESTCD = if_else(ItemOID == "IT.LB.CALCIUM", "CA", "CREAT"),32    LBTEST = if_else(ItemOID == "IT.LB.CALCIUM", "Calcium", "Creatinine"),33    NOTDONE = Value == "NOT DONE",34    DOMAIN = "LB",35    STUDYID = StudyOID,36    USUBJID = SubjectKey,37    LBORRES = if_else(NOTDONE, NA_character_, Value),38    LBORRESU = if_else(NOTDONE, NA_character_, "mg/dL"),39    LBSTRESN = if_else(40      NOTDONE, NA_real_, suppressWarnings(as.numeric(Value))41    ),42    LBSTRESU = if_else(NOTDONE, NA_character_, "mg/dL"),43    LBSTAT = if_else(NOTDONE, "NOT DONE", NA_character_)44  ) |>45  arrange(STUDYID, USUBJID, LBDTC, LBTESTCD) |>46  group_by(STUDYID, USUBJID) |>47  mutate(LBSEQ = row_number()) |>48  ungroup() |>49  select(50    DOMAIN, STUDYID, USUBJID, LBSEQ, LBTESTCD, LBTEST, LBORRES,51    LBORRESU, LBSTRESN, LBSTRESU, LBSTAT, LBDTC52  )5354dir.create("/app/output", showWarnings = FALSE)55write_csv(lb, "/app/output/lb.csv", na = "")

1# Reference solution for the yamaa benchmark sdtm-lb-findings (Python 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 1.67"""Build the SDTM LB dataset from the ODM extract.89One record per collected lab result in the IG.LB item groups: a test10with no collected entry gives no record, while an entry of "NOT DONE"11gives a record carrying LBSTAT.12"""1314import csv15import os1617INPUT_DIR = "/app/input"18OUTPUT_DIR = "/app/output"1920TESTS = {21    "IT.LB.CALCIUM": ("CA", "Calcium"),22    "IT.LB.CREAT": ("CREAT", "Creatinine"),23}24DATE_ITEM = "IT.LB.LBDTC"25UNIT = "mg/dL"2627COLUMNS = [28    "DOMAIN",29    "STUDYID",30    "USUBJID",31    "LBSEQ",32    "LBTESTCD",33    "LBTEST",34    "LBORRES",35    "LBORRESU",36    "LBSTRESN",37    "LBSTRESU",38    "LBSTAT",39    "LBDTC",40]414243def numeric_text(value):44    """Numeric form of a collected entry, or "" when it is not a number."""45    try:46        number = float(value)47    except ValueError:48        return ""49    if number.is_integer():50        return str(int(number))51    return str(number)525354def main():55    with open(os.path.join(INPUT_DIR, "odm.csv"), newline="") as handle:56        rows = list(csv.DictReader(handle))5758    # Group the LB items by subject and visit group.59    groups = {}60    for row in rows:61        if row["ItemGroupOID"] != "IG.LB":62            continue63        key = (64            row["StudyOID"],65            row["SubjectKey"],66            row["StudyEventOID"],67            row["StudyEventRepeatKey"],68            row["ItemGroupOID"],69            row["ItemGroupRepeatKey"],70        )71        groups.setdefault(key, {})[row["ItemOID"]] = row["Value"]7273    records = []74    for (study_id, subject, *_), items in groups.items():75        lbdtc = items.get(DATE_ITEM, "")76        for item_oid, (testcd, test) in TESTS.items():77            entry = items.get(item_oid)78            if entry is None or entry == "":79                continue  # no collected entry: no record80            record = {81                "DOMAIN": "LB",82                "STUDYID": study_id,83                "USUBJID": subject,84                "LBTESTCD": testcd,85                "LBTEST": test,86                "LBDTC": lbdtc,87            }88            if entry == "NOT DONE":89                record.update(90                    LBORRES="",91                    LBORRESU="",92                    LBSTRESN="",93                    LBSTRESU="",94                    LBSTAT="NOT DONE",95                )96            else:97                record.update(98                    LBORRES=entry,99                    LBORRESU=UNIT,100                    LBSTRESN=numeric_text(entry),101                    LBSTRESU=UNIT,102                    LBSTAT="",103                )104            records.append(record)105106    # LBSEQ numbers each subject's records from 1 by collection date,107    # then by test code.108    by_subject = {}109    for record in records:110        by_subject.setdefault(record["USUBJID"], []).append(record)111    ordered = []112    for subject in sorted(by_subject):113        subject_records = sorted(114            by_subject[subject],115            key=lambda r: (r["LBDTC"], r["LBTESTCD"]),116        )117        for index, record in enumerate(subject_records, start=1):118            record["LBSEQ"] = str(index)119            ordered.append(record)120121    os.makedirs(OUTPUT_DIR, exist_ok=True)122    with open(os.path.join(OUTPUT_DIR, "lb.csv"), "w", newline="") as handle:123        writer = csv.DictWriter(handle, fieldnames=COLUMNS)124        writer.writeheader()125        writer.writerows(ordered)126127128if __name__ == "__main__":129    main()

1/* opensas reference solution for the yamaa benchmark sdtm-lb-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 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. ItemGroupOID : $char1024. ItemGroupRepeatKey : $char1024. ItemOID : $char1024. Value : $char1024.;9run;1011proc sort data=odm out=ordered; by StudyOID SubjectKey StudyEventOID StudyEventRepeatKey ItemGroupRepeatKey; run;12data dates;13  length LBDTC $1024;14  retain LBDTC;15  set ordered;1617  by StudyOID SubjectKey StudyEventOID StudyEventRepeatKey ItemGroupRepeatKey;18  if first.ItemGroupRepeatKey then do;19    LBDTC='';20  end;21    if ItemOID='IT.LB.LBDTC' then LBDTC=Value;22  if last.ItemGroupRepeatKey then output;23run;24proc sql;25create table joined as select a.*,b.LBDTC from odm a inner join dates b on a.StudyOID=b.StudyOID and a.SubjectKey=b.SubjectKey and a.StudyEventOID=b.StudyEventOID and a.StudyEventRepeatKey=b.StudyEventRepeatKey and a.ItemGroupRepeatKey=b.ItemGroupRepeatKey where a.ItemOID in ('IT.LB.CALCIUM','IT.LB.CREAT') and not missing(a.Value); quit;26data prepared; set joined; length DOMAIN $2 STUDYID USUBJID LBTESTCD LBTEST LBORRES LBORRESU LBSTRESU LBSTAT $1024;27DOMAIN='LB'; STUDYID=StudyOID; USUBJID=SubjectKey; LBORRES=Value; LBORRESU='mg/dL'; LBSTRESU='mg/dL'; LBSTRESN=.; LBSTAT='';28if Value ne 'NOT DONE' then LBSTRESN=input(Value,?? best32.);29if ItemOID='IT.LB.CREAT' then do; LBTESTCD='CREAT'; LBTEST='Creatinine'; end; else do; LBTESTCD='CA'; LBTEST='Calcium'; end;30if Value='NOT DONE' then do; LBORRES=''; LBORRESU=''; LBSTRESU=''; LBSTRESN=.; LBSTAT='NOT DONE'; end; run;31proc sort data=prepared; by STUDYID USUBJID LBDTC LBTESTCD; run;32data result; set prepared; by STUDYID USUBJID; retain LBSEQ;33  if first.USUBJID then LBSEQ=0;34  LBSEQ+1;35run;3637proc sql;38  create table final as select DOMAIN, STUDYID, USUBJID, LBSEQ, LBTESTCD, LBTEST, LBORRES, LBORRESU, LBSTRESN, LBSTRESU, LBSTAT, LBDTC from result;39quit;40proc export data=final outfile="/app/output/lb.csv" dbms=csv replace; run;