Multiple Findings About from ODM

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

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

1# Reference solution for the yamaa benchmark sdtm-fa-odm-multitest (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(purrr)1011odm <- read_csv(12  "/app/input/odm.csv",13  col_types = cols(.default = col_character())14) |>15  mutate(Value = na_if(Value, ""))1617forms <- odm |>18  filter(ItemGroupOID == "IG.FA") |>19  distinct(20    StudyOID, SubjectKey, StudyEventOID, StudyEventRepeatKey,21    ItemGroupRepeatKey22  )2324one_form <- function(study, subj, event, rep, igrep) {25  sub <- odm |>26    filter(27      SubjectKey == subj,28      StudyEventOID == event,29      StudyEventRepeatKey == rep,30      ItemGroupOID == "IG.FA",31      ItemGroupRepeatKey == igrep32    )33  present <- sub$ItemOID34  get <- function(oid) {35    hit <- sub |> filter(ItemOID == oid)36    if (nrow(hit) == 0) return(list(present = FALSE, value = NA_character_))37    list(present = TRUE, value = hit$Value[[1]])38  }39  obj <- get("IT.FA.FAOBJ")40  if (!obj$present || is.na(obj$value)) return(tibble())41  fadtc <- get("IT.FA.FADTC")$value42  day <- dplyr::recode(event, DAY1 = 1L, DAY2 = 2L, .default = 99L)43  fatpt <- dplyr::recode(44    event, DAY1 = "END DAY 1", DAY2 = "END DAY 2", .default = NA_character_45  )46  specs <- list(47    list(oid = "IT.FA.OCCUR", code = "OCCUR", name = "Occurrence Indicator"),48    list(oid = "IT.FA.SEV", code = "SEV", name = "Severity/Intensity"),49    list(oid = "IT.FA.LDIAM", code = "LDIAM", name = "Longest Diameter")50  )51  imap_dfr(specs, function(spec, i) {52    g <- get(spec$oid)53    if (!g$present) return(tibble())54    blank <- is.na(g$value)55    tibble(56      DOMAIN = "FA",57      STUDYID = study,58      USUBJID = subj,59      DAY = day,60      REP = as.integer(igrep),61      ORDER = as.integer(i - 1L),62      FATESTCD = spec$code,63      FATEST = spec$name,64      FAOBJ = obj$value,65      FACAT = "REACTOGENICITY",66      FAORRES = g$value,67      FAORRESU = if_else(spec$oid == "IT.FA.LDIAM", "mm", NA_character_),68      FASTRESC = g$value,69      FASTAT = if_else(blank, "NOT DONE", NA_character_),70      FATPT = fatpt,71      FADTC = fadtc72    )73  })74}7576fa <- pmap_dfr(77  list(78    forms$StudyOID, forms$SubjectKey, forms$StudyEventOID,79    forms$StudyEventRepeatKey, forms$ItemGroupRepeatKey80  ),81  one_form82) |>83  arrange(STUDYID, USUBJID, DAY, REP, ORDER) |>84  group_by(STUDYID, USUBJID) |>85  mutate(FASEQ = row_number()) |>86  ungroup() |>87  select(88    DOMAIN, STUDYID, USUBJID, FASEQ, FATESTCD, FATEST, FAOBJ, FACAT,89    FAORRES, FAORRESU, FASTRESC, FASTAT, FATPT, FADTC90  )9192dir.create("/app/output", showWarnings = FALSE)93write_csv(fa, "/app/output/fa.csv", na = "")

1# Reference solution for the yamaa benchmark sdtm-fa-odm-multitest (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 FA (Findings About) dataset from the ODM extract.89One record per collected reaction finding: for each diary occurrence10(an IG.FA item group repeat within a study event), one record for each11of the occurrence, severity, and longest diameter result items that was12collected. A result item present with a blank result still gives its13record, flagged FASTAT = "NOT DONE".14"""1516import csv17import re1819INPUT_ODM = "/app/input/odm.csv"20OUTPUT_FA = "/app/output/fa.csv"2122COLUMNS = [23    "DOMAIN", "STUDYID", "USUBJID", "FASEQ", "FATESTCD", "FATEST",24    "FAOBJ", "FACAT", "FAORRES", "FAORRESU", "FASTRESC", "FASTAT",25    "FATPT", "FADTC",26]2728# Result items in test order: (item OID suffix, FATESTCD, FATEST).29TESTS = [30    ("OCCUR", "OCCUR", "Occurrence Indicator"),31    ("SEV", "SEV", "Severity/Intensity"),32    ("LDIAM", "LDIAM", "Longest Diameter"),33]343536def day_number(study_event_oid):37    """Numeric part of a diary day OID such as DAY1 or DAY2."""38    match = re.search(r"(\d+)\s*$", study_event_oid)39    return int(match.group(1)) if match else 0404142def main():43    with open(INPUT_ODM, newline="", encoding="utf-8") as handle:44        rows = list(csv.DictReader(handle))4546    # Group the IG.FA items by diary occurrence: subject, study event47    # (diary day) and its repeat, and the item group (reaction) repeat.48    occurrences = {}49    subject_order = []50    for row in rows:51        if row["ItemGroupOID"] != "IG.FA":52            continue53        key = (54            row["SubjectKey"],55            row["StudyEventOID"],56            row["StudyEventRepeatKey"],57            row["ItemGroupRepeatKey"],58        )59        if key not in occurrences:60            occurrences[key] = {"study": row["StudyOID"], "items": {}}61            if row["SubjectKey"] not in subject_order:62                subject_order.append(row["SubjectKey"])63        item = row["ItemOID"].rsplit(".", 1)[-1]64        occurrences[key]["items"][item] = row["Value"]6566    records_by_subject = {}67    for (subject, event_oid, event_repeat, group_repeat), occ in occurrences.items():68        items = occ["items"]69        reaction = items.get("FAOBJ", "")70        fadtc = items.get("FADTC", "")71        day = day_number(event_oid)72        fatpt = "END DAY %d" % day73        for test_index, (item_key, testcd, test) in enumerate(TESTS):74            if item_key not in items:75                # The result item was not collected for this reaction:76                # no record.77                continue78            result = items[item_key]79            record = {80                "DOMAIN": "FA",81                "STUDYID": occ["study"],82                "USUBJID": subject,83                "FATESTCD": testcd,84                "FATEST": test,85                "FAOBJ": reaction,86                "FACAT": "REACTOGENICITY",87                "FAORRES": result,88                "FAORRESU": "mm" if testcd == "LDIAM" else "",89                "FASTRESC": result,90                "FASTAT": "NOT DONE" if result == "" else "",91                "FATPT": fatpt,92                "FADTC": fadtc,93            }94            sort_key = (day, int(event_repeat), int(group_repeat), test_index)95            records_by_subject.setdefault(subject, []).append((sort_key, record))9697    out_rows = []98    for subject in subject_order:99        subject_records = sorted(100            records_by_subject.get(subject, []), key=lambda pair: pair[0]101        )102        for seq, (_, record) in enumerate(subject_records, start=1):103            record["FASEQ"] = str(seq)104            out_rows.append(record)105106    with open(OUTPUT_FA, "w", newline="", encoding="utf-8") as handle:107        writer = csv.DictWriter(handle, fieldnames=COLUMNS)108        writer.writeheader()109        writer.writerows(out_rows)110111112if __name__ == "__main__":113    main()

1/* opensas reference solution for the yamaa benchmark sdtm-fa-odm-multitest.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 siblings;13  length FAOBJ FADTC $1024;14  retain FAOBJ FADTC;15  set ordered;16  where ItemGroupOID='IG.FA';17  by StudyOID SubjectKey StudyEventOID StudyEventRepeatKey ItemGroupRepeatKey;18  if first.ItemGroupRepeatKey then do;19    FAOBJ=''; FADTC='';20  end;21    if ItemOID='IT.FA.FAOBJ' then FAOBJ=Value;22  if ItemOID='IT.FA.FADTC' then FADTC=Value;23  if last.ItemGroupRepeatKey then output;24run;25proc sql;26create table joined as select a.*,b.FAOBJ,b.FADTC from odm a inner join siblings 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.FA.OCCUR','IT.FA.SEV','IT.FA.LDIAM'); quit;27data prepared; set joined; length DOMAIN $2 STUDYID USUBJID FATESTCD FATEST FACAT FAORRES FAORRESU FASTRESC FASTAT FATPT $1024;28DOMAIN='FA'; STUDYID=StudyOID; USUBJID=SubjectKey; FACAT='REACTOGENICITY'; FAORRES=Value; FASTRESC=Value; FAORRESU=''; FASTAT=''; if missing(Value) then FASTAT='NOT DONE';29if ItemOID='IT.FA.OCCUR' then do; FATESTCD='OCCUR'; FATEST='Occurrence Indicator'; TESTORD=1; end;30if ItemOID='IT.FA.SEV' then do; FATESTCD='SEV'; FATEST='Severity/Intensity'; TESTORD=2; end;31if ItemOID='IT.FA.LDIAM' then do; FATESTCD='LDIAM'; FATEST='Longest Diameter'; FAORRESU='mm'; TESTORD=3; end;32FATPT='END DAY 1'; if StudyEventOID='DAY2' then FATPT='END DAY 2'; DAYORD=input(compress(StudyEventOID,,'kd'),best32.); REP=input(ItemGroupRepeatKey,best32.); run;33proc sort data=prepared; by STUDYID USUBJID DAYORD REP TESTORD; run;34data result; set prepared; by STUDYID USUBJID; retain FASEQ;35  if first.USUBJID then FASEQ=0;36  FASEQ+1;37run;3839proc sql;40  create table final as select DOMAIN, STUDYID, USUBJID, FASEQ, FATESTCD, FATEST, FAOBJ, FACAT, FAORRES, FAORRESU, FASTRESC, FASTAT, FATPT, FADTC from result;41quit;42proc export data=final outfile="/app/output/fa.csv" dbms=csv replace; run;