Pre-specified Medical History Checklist

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

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

1# Reference solution for the yamaa benchmark sdtm-mh-prespecified-conditions (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)1011items <- read_csv(12  "/app/input/mh_items.csv",13  col_types = cols(SORTORD = col_integer(), .default = col_character())14)15odm <- read_csv(16  "/app/input/odm.csv",17  col_types = cols(.default = col_character())18)1920checklist_oids <- intersect(items$ItemOID, unique(odm$ItemOID))2122checklist <- odm |>23  filter(ItemOID %in% checklist_oids) |>24  group_by(StudyOID, SubjectKey, ItemOID) |>25  slice_tail(n = 1) |>26  ungroup() |>27  left_join(items |> select(ItemOID, MHTERM, SORTORD), by = "ItemOID") |>28  mutate(29    answered = Value %in% c("Y", "N"),30    MHTERM = MHTERM,31    MHCAT = "DISEASE-SPECIFIC HISTORY",32    MHPRESP = "Y",33    MHOCCUR = if_else(answered, Value, NA_character_),34    MHSTAT = if_else(answered, NA_character_, "NOT DONE"),35    visit_rank = 0L,36    repeat_no = 0L37  ) |>38  select(39    STUDYID = StudyOID, USUBJID = SubjectKey, ItemOID, MHTERM, MHCAT,40    MHPRESP, MHOCCUR, MHSTAT, SORTORD, visit_rank, repeat_no41  )4243volunteered <- odm |>44  filter(!ItemOID %in% checklist_oids, !is.na(Value) & Value != "") |>45  mutate(46    MHTERM = Value,47    MHCAT = "GENERAL HISTORY",48    MHPRESP = NA_character_,49    MHOCCUR = NA_character_,50    MHSTAT = NA_character_,51    SORTORD = NA_integer_,52    visit_rank = case_when(53      StudyEventOID == "SCREENING" ~ 0L,54      StudyEventOID == "BASELINE" ~ 1L,55      TRUE ~ 99L56    ),57    repeat_no = suppressWarnings(as.integer(ItemGroupRepeatKey))58  ) |>59  select(60    STUDYID = StudyOID, USUBJID = SubjectKey, ItemOID, MHTERM, MHCAT,61    MHPRESP, MHOCCUR, MHSTAT, SORTORD, visit_rank, repeat_no62  )6364mh <- bind_rows(checklist, volunteered) |>65  arrange(USUBJID, is.na(SORTORD), SORTORD, visit_rank, repeat_no) |>66  group_by(USUBJID) |>67  mutate(MHSEQ = row_number()) |>68  ungroup() |>69  mutate(DOMAIN = "MH") |>70  arrange(USUBJID, MHSEQ) |>71  select(72    DOMAIN, STUDYID, USUBJID, MHSEQ, MHTERM, MHCAT, MHPRESP, MHOCCUR, MHSTAT73  )7475dir.create("/app/output", showWarnings = FALSE)76write_csv(mh, "/app/output/mh.csv", na = "")

1# Reference solution for the yamaa benchmark sdtm-mh-prespecified-conditions (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 1.67from pathlib import Path89import polars as pl101112def _present(value) -> bool:13    return value is not None and str(value) != ""141516items = pl.read_csv("/app/input/mh_items.csv", infer_schema=False)17odm = pl.read_csv("/app/input/odm.csv", infer_schema=False)1819meta = {row["ItemOID"]: (row["MHTERM"], int(row["SORTORD"])) for row in items.to_dicts()}20checklist_oids = {oid for oid in meta if (odm["ItemOID"] == oid).any()}2122visit_order = {"SCREENING": 0, "BASELINE": 1}2324records_by_subject: dict[tuple[str, str], dict] = {}25for row in odm.to_dicts():26    studyid, usubjid = str(row["StudyOID"]), str(row["SubjectKey"])27    key = (studyid, usubjid)28    records_by_subject.setdefault(key, {"checklist": {}, "volunteered": []})2930for row in odm.to_dicts():31    key = (str(row["StudyOID"]), str(row["SubjectKey"]))32    oid, value = str(row["ItemOID"]), row["Value"]33    if oid in checklist_oids:34        # Last collected answer wins when an item repeats, including blanks.35        records_by_subject[key]["checklist"][oid] = value if _present(value) else None36    elif not _present(value):37        continue38    else:39        # Each volunteered condition is its own record, even when its40        # repeat number is reused at another visit.41        event = str(row["StudyEventOID"])42        records_by_subject[key]["volunteered"].append(43            {44                "term": str(value),45                "visit_rank": visit_order.get(event, 99),46                "repeat": int(row["ItemGroupRepeatKey"] or 1),47            }48        )4950rows = []51for (studyid, usubjid), parts in records_by_subject.items():52    ordered = sorted(parts["checklist"].items(), key=lambda kv: meta[kv[0]][1])53    seq = 054    for oid, answer in ordered:55        seq += 156        term, _ = meta[oid]57        answered = answer in ("Y", "N")58        rows.append(59            {60                "DOMAIN": "MH",61                "STUDYID": studyid,62                "USUBJID": usubjid,63                "MHSEQ": str(seq),64                "MHTERM": term,65                "MHCAT": "DISEASE-SPECIFIC HISTORY",66                "MHPRESP": "Y",67                "MHOCCUR": answer if answered else None,68                "MHSTAT": None if answered else "NOT DONE",69            }70        )71    volunteered = sorted(72        parts["volunteered"], key=lambda v: (v["visit_rank"], v["repeat"])73    )74    for entry in volunteered:75        seq += 176        rows.append(77            {78                "DOMAIN": "MH",79                "STUDYID": studyid,80                "USUBJID": usubjid,81                "MHSEQ": str(seq),82                "MHTERM": entry["term"],83                "MHCAT": "GENERAL HISTORY",84                "MHPRESP": None,85                "MHOCCUR": None,86                "MHSTAT": None,87            }88        )8990mh = (91    pl.DataFrame(92        rows,93        schema={94            "DOMAIN": pl.String,95            "STUDYID": pl.String,96            "USUBJID": pl.String,97            "MHSEQ": pl.String,98            "MHTERM": pl.String,99            "MHCAT": pl.String,100            "MHPRESP": pl.String,101            "MHOCCUR": pl.String,102            "MHSTAT": pl.String,103        },104    )105    .with_columns(pl.col("MHSEQ").cast(pl.Int64).alias("_seq"))106    .sort(["USUBJID", "_seq"])107    .select(108        "DOMAIN", "STUDYID", "USUBJID", "MHSEQ", "MHTERM", "MHCAT",109        "MHPRESP", "MHOCCUR", "MHSTAT",110    )111)112113Path("/app/output").mkdir(exist_ok=True)114mh.write_csv("/app/output/mh.csv")

1/* opensas reference solution for the yamaa benchmark sdtm-mh-prespecified-conditions.2   Reads the benchmark inputs and derives the requested output.3   Execute with the existing opensas runtime: opensas /app/output/result.sas. */45data mh_items;6  length ItemOID MHTERM SORTORD $1024;7  infile "/app/input/mh_items.csv" dsd dlm="," firstobs=2 truncover lrecl=32767;8  input ItemOID : $char1024. MHTERM : $char1024. SORTORD : $char1024.;9run;1011data odm;12  length StudyOID MetaDataVersionOID SubjectKey StudyEventOID StudyEventRepeatKey ItemGroupOID ItemGroupRepeatKey ItemOID Value $1024;13  infile "/app/input/odm.csv" dsd dlm="," firstobs=2 truncover lrecl=32767;14  input StudyOID : $char1024. MetaDataVersionOID : $char1024. SubjectKey : $char1024. StudyEventOID : $char1024. StudyEventRepeatKey : $char1024. ItemGroupOID : $char1024. ItemGroupRepeatKey : $char1024. ItemOID : $char1024. Value : $char1024.;15run;1617data authored; set odm; ordinal=_n_; run;18proc sql; create table checklist as select a.*,b.MHTERM,b.SORTORD from authored a inner join mh_items b on a.ItemOID=b.ItemOID; quit;19proc sort data=checklist; by StudyOID SubjectKey ItemOID ordinal; run;20data checklistlast; set checklist; by StudyOID SubjectKey ItemOID; if last.ItemOID; run;21data checklistout; set checklistlast; length DOMAIN $2 STUDYID USUBJID MHCAT MHPRESP MHOCCUR MHSTAT $1024; DOMAIN='MH'; STUDYID=StudyOID; USUBJID=SubjectKey; MHCAT='DISEASE-SPECIFIC HISTORY'; MHPRESP='Y'; MHOCCUR=''; MHSTAT='NOT DONE'; if Value in ('Y','N') then do; MHOCCUR=Value; MHSTAT=''; end; SECTION=0; ORD=input(SORTORD,best32.); VISITORD=0; REP=0; run;22data volunteered; set odm; where ItemOID='IT.MH.MHTERM' and not missing(Value); length DOMAIN $2 STUDYID USUBJID MHTERM MHCAT MHPRESP MHOCCUR MHSTAT $1024; DOMAIN='MH'; STUDYID=StudyOID; USUBJID=SubjectKey; MHTERM=Value; MHCAT='GENERAL HISTORY'; MHPRESP=''; MHOCCUR=''; MHSTAT=''; SECTION=1; ORD=0; VISITORD=0; if StudyEventOID='BASELINE' then VISITORD=1; REP=input(ItemGroupRepeatKey,best32.); run;23data prepared; set checklistout volunteered; run;24proc sort data=prepared; by STUDYID USUBJID SECTION ORD VISITORD REP; run;25data result; set prepared; by STUDYID USUBJID; retain MHSEQ;26  if first.USUBJID then MHSEQ=0;27  MHSEQ+1;28run;2930proc sql;31  create table final as select DOMAIN, STUDYID, USUBJID, MHSEQ, MHTERM, MHCAT, MHPRESP, MHOCCUR, MHSTAT from result;32quit;33proc export data=final outfile="/app/output/mh.csv" dbms=csv replace; run;