Replicate Readings

1library(yamaa)23vs <- yamaa_domain("spec.yaml")$output4vs

1import yamaa23vs = yamaa.yamaa_domain("spec.yaml").output4vs

1# Reference solution for the yamaa benchmark sdtm-vs-replicates (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)1011fmt_text <- function(x) {12  if (is.na(x)) {13    return(NA_character_)14  }15  if (x == floor(x)) {16    formatC(x, format = "f", digits = 0)17  } else {18    format(x, scientific = FALSE, trim = TRUE, digits = 15)19  }20}2122odm <- read_csv(23  "/app/input/odm.csv",24  col_types = cols(.default = col_character())25)2627readings <- odm |>28  filter(29    ItemOID %in% c("IT.VS.SYSBP", "IT.VS.DIABP"),30    StudyEventOID %in% c("SE.VISIT1", "SE.VISIT2"),31    !is.na(Value) & Value != ""32  ) |>33  mutate(34    VISIT = case_when(35      StudyEventOID == "SE.VISIT1" ~ "VISIT 1",36      .default = "VISIT 2"37    ),38    VSTESTCD = case_when(39      ItemOID == "IT.VS.DIABP" ~ "DIABP",40      .default = "SYSBP"41    ),42    VSTEST = case_when(43      ItemOID == "IT.VS.DIABP" ~ "Diastolic Blood Pressure",44      .default = "Systolic Blood Pressure"45    ),46    VSREPNUM = suppressWarnings(as.integer(ItemGroupRepeatKey)),47    val = suppressWarnings(as.numeric(Value)),48    STUDYID = StudyOID,49    USUBJID = SubjectKey50  ) |>51  filter(!is.na(val), !is.na(VSREPNUM)) |>52  select(STUDYID, USUBJID, VISIT, VSTESTCD, VSTEST, VSREPNUM, val)5354mean_records <- readings |>55  group_by(STUDYID, USUBJID, VISIT, VSTESTCD, VSTEST) |>56  summarise(mean_val = mean(val), .groups = "drop")5758vs <- bind_rows(59  readings |>60    transmute(61      DOMAIN = "VS",62      STUDYID, USUBJID, VISIT, VSTESTCD, VSTEST,63      VSREPNUM,64      VSORRES = vapply(val, fmt_text, character(1)),65      VSORRESU = "mmHg",66      VSSTRESN = val,67      VSSTRESC = VSORRES,68      VSDRVFL = NA_character_69    ),70  mean_records |>71    transmute(72      DOMAIN = "VS",73      STUDYID, USUBJID, VISIT, VSTESTCD, VSTEST,74      VSREPNUM = NA_integer_,75      VSORRES = vapply(mean_val, fmt_text, character(1)),76      VSORRESU = "mmHg",77      VSSTRESN = mean_val,78      VSSTRESC = VSORRES,79      VSDRVFL = "Y"80    )81) |>82  arrange(USUBJID, VISIT, VSTESTCD, is.na(VSREPNUM), VSREPNUM) |>83  group_by(USUBJID) |>84  mutate(VSSEQ = row_number()) |>85  ungroup() |>86  arrange(USUBJID, VSSEQ) |>87  select(88    DOMAIN, STUDYID, USUBJID, VSSEQ, VISIT, VSTESTCD, VSTEST, VSREPNUM,89    VSORRES, VSORRESU, VSSTRESN, VSSTRESC, VSDRVFL90  )9192dir.create("/app/output", showWarnings = FALSE)93write_csv(vs, "/app/output/vs.csv", na = "")

1# Reference solution for the yamaa benchmark sdtm-vs-replicates (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 fmt_text(x: float | None) -> str | None:13    if x is None:14        return None15    import math1617    if isinstance(x, float) and (math.isnan(x) or math.isinf(x)):18        return None19    if float(x).is_integer():20        return str(int(float(x)))21    return f"{float(x):.15g}"222324odm = pl.read_csv("/app/input/odm.csv", infer_schema=False)2526test_name = {27    "IT.VS.SYSBP": ("SYSBP", "Systolic Blood Pressure"),28    "IT.VS.DIABP": ("DIABP", "Diastolic Blood Pressure"),29}30visit_name = {"SE.VISIT1": "VISIT 1", "SE.VISIT2": "VISIT 2"}3132# Collected readings: one per subject, visit, test, and replicate. A blank33# value or an absent record gives no reading record.34readings: dict[tuple[str, str, str, str, str], float] = {}35studies: dict[str, str] = {}36for row in odm.to_dicts():37    if row["ItemOID"] not in test_name or row["StudyEventOID"] not in visit_name:38        continue39    value = row["Value"]40    if value is None or value == "":41        continue42    studies[row["SubjectKey"]] = row["StudyOID"]43    key = (44        row["SubjectKey"],45        row["StudyEventOID"],46        row["ItemOID"],47        row["ItemGroupRepeatKey"],48    )49    readings[key] = float(value)50    testcd, _ = test_name[row["ItemOID"]]51    _ = testcd5253records = []54# By visit name, then DIABP before SYSBP; each test's readings in replicate55# order, then its mean record.56for subject in sorted(studies):57    for event in sorted(visit_name, key=lambda e: visit_name[e]):58        visit = visit_name[event]59        for item in ("IT.VS.DIABP", "IT.VS.SYSBP"):60            testcd, test = test_name[item]61            reps = sorted(62                (rep, val)63                for (s, e, i, rep), val in readings.items()64                if s == subject and e == event and i == item65            )66            if not reps:67                continue68            for rep, val in reps:69                records.append(70                    {71                        "DOMAIN": "VS",72                        "STUDYID": studies[subject],73                        "USUBJID": subject,74                        "VISIT": visit,75                        "VSTESTCD": testcd,76                        "VSTEST": test,77                        "VSREPNUM": int(rep),78                        "VSORRES": fmt_text(val),79                        "VSORRESU": "mmHg",80                        "VSSTRESN": val,81                        "VSSTRESC": fmt_text(val),82                        "VSDRVFL": None,83                        "_visit_ord": visit,84                        "_test_ord": testcd,85                        "_rep_ord": int(rep),86                        "_mean_last": 0,87                    }88                )89            mean = sum(val for _, val in reps) / len(reps)90            records.append(91                {92                    "DOMAIN": "VS",93                    "STUDYID": studies[subject],94                    "USUBJID": subject,95                    "VISIT": visit,96                    "VSTESTCD": testcd,97                    "VSTEST": test,98                    "VSREPNUM": None,99                    "VSORRES": fmt_text(mean),100                    "VSORRESU": "mmHg",101                    "VSSTRESN": mean,102                    "VSSTRESC": fmt_text(mean),103                    "VSDRVFL": "Y",104                    "_visit_ord": visit,105                    "_test_ord": testcd,106                    "_rep_ord": 999,107                    "_mean_last": 1,108                }109            )110111vs = (112    pl.DataFrame(records)113    .sort(["USUBJID", "_visit_ord", "_test_ord", "_mean_last", "_rep_ord"])114    .with_columns(VSSEQ=pl.int_range(1, pl.len() + 1).over("USUBJID"))115    .sort(["USUBJID", "VSSEQ"])116    .select(117        "DOMAIN",118        "STUDYID",119        "USUBJID",120        "VSSEQ",121        "VISIT",122        "VSTESTCD",123        "VSTEST",124        "VSREPNUM",125        "VSORRES",126        "VSORRESU",127        "VSSTRESN",128        "VSSTRESC",129        "VSDRVFL",130    )131)132133Path("/app/output").mkdir(exist_ok=True)134vs.write_csv("/app/output/vs.csv")

1/* opensas reference solution for the yamaa benchmark sdtm-vs-replicates.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;1011data readings; set odm; where ItemOID in ('IT.VS.SYSBP','IT.VS.DIABP') and not missing(Value); length STUDYID USUBJID VISIT VSTESTCD $1024;12STUDYID=StudyOID; USUBJID=SubjectKey; VISIT=catx(' ','VISIT',scan(StudyEventOID,-1,'T')); if StudyEventOID='SE.VISIT1' then VISIT='VISIT 1'; if StudyEventOID='SE.VISIT2' then VISIT='VISIT 2'; VSTESTCD=scan(ItemOID,-1,'.'); VSREPNUM=input(ItemGroupRepeatKey,best32.); VSSTRESN=input(Value,best32.); MEANORD=0; run;13proc sql; create table means as select STUDYID,USUBJID,VISIT,VSTESTCD,mean(VSSTRESN) as VSSTRESN from readings group by STUDYID,USUBJID,VISIT,VSTESTCD; quit;14data meanout; set means; VSREPNUM=.; MEANORD=1; run;15data prepared; set readings meanout; length DOMAIN $2 VSTEST VSORRES VSORRESU VSSTRESC VSDRVFL $1024; DOMAIN='VS'; VSTEST='Diastolic Blood Pressure'; if VSTESTCD='SYSBP' then VSTEST='Systolic Blood Pressure'; VSORRES=strip(put(VSSTRESN,best32.)); VSSTRESC=VSORRES; VSORRESU='mmHg'; VSDRVFL=''; if MEANORD=1 then VSDRVFL='Y'; run;16proc sort data=prepared; by STUDYID USUBJID VISIT VSTESTCD MEANORD VSREPNUM; run;17data result; set prepared; by STUDYID USUBJID; retain VSSEQ;18  if first.USUBJID then VSSEQ=0;19  VSSEQ+1;20run;2122proc sql;23  create table final as select DOMAIN, STUDYID, USUBJID, VSSEQ, VISIT, VSTESTCD, VSTEST, VSREPNUM, VSORRES, VSORRESU, VSSTRESN, VSSTRESC, VSDRVFL from result;24quit;25proc export data=final outfile="/app/output/vs.csv" dbms=csv replace; run;