Uncollected Exposure

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

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

1# Reference solution for the yamaa benchmark adam-adex-uncollected (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)910dm <- read_csv(11  "/app/input/dm.csv",12  col_types = cols(.default = col_character())13)14ex <- read_csv(15  "/app/input/ex.csv",16  col_types = cols(EXDOSE = col_double(), .default = col_character())17)1819if (any(!is.na(ex$EXDOSE) & (is.na(ex$EXDOSU) | ex$EXDOSU != "mg"))) {20  stop("Recorded doses must be standardized to mg")21}2223# Per subject: total dose, record count, and count of recorded doses.24agg <- ex |>25  group_by(STUDYID, USUBJID) |>26  summarise(27    DOSE_SUM = sum(EXDOSE, na.rm = TRUE),28    NDOSREC = n(),29    NDOSVAL = sum(!is.na(EXDOSE)),30    .groups = "drop"31  ) |>32  mutate(DOSECUM = if_else(NDOSVAL > 0, DOSE_SUM, NA_real_)) |>33  select(STUDYID, USUBJID, DOSECUM, NDOSREC, NDOSVAL)3435adex <- dm |>36  left_join(agg, by = c("STUDYID", "USUBJID")) |>37  transmute(STUDYID, USUBJID, TRT01P = ARM, DOSECUM, NDOSREC, NDOSVAL)3839dir.create("/app/output", showWarnings = FALSE)40write_csv(adex, "/app/output/adex.csv", na = "")

1# Reference solution for the yamaa benchmark adam-adex-uncollected (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"""Create the ADEX dataset from SDTM DM and EX.89One record per DM subject, carrying ARM as the planned treatment TRT01P.10DOSECUM is the total of the recorded EXDOSE values on the subject's EX11records; it has no value when no dose was ever recorded, while a12recorded total of zero is kept as zero. NDOSREC is the number of the13subject's EX records, with no value when the subject has no EX record at14all. NDOSVAL is the number of those EX records that carry a recorded15dose; an explicitly recorded zero counts, and NDOSVAL is 0 when the16subject has EX records but every dose was left blank.17"""1819import csv20from collections import defaultdict2122INPUT_DIR = "/app/input"23OUTPUT_PATH = "/app/output/adex.csv"242526def read_rows(path):27    with open(path, newline="", encoding="utf-8") as handle:28        return list(csv.DictReader(handle))293031def format_number(value):32    """Write a number without a trailing decimal part when whole."""33    if value == int(value):34        return str(int(value))35    return repr(value)363738def main():39    subjects = read_rows(f"{INPUT_DIR}/dm.csv")40    exposure = read_rows(f"{INPUT_DIR}/ex.csv")4142    doses_by_key = defaultdict(list)43    for row in exposure:44        if row["EXDOSE"].strip() and row["EXDOSU"] != "mg":45            raise ValueError("Recorded doses must be standardized to mg")46        key = (row["STUDYID"], row["USUBJID"])47        doses_by_key[key].append(row["EXDOSE"])4849    records = []50    for row in subjects:51        key = (row["STUDYID"], row["USUBJID"])52        doses = doses_by_key.get(key)53        if doses is None:54            n_records = ""55            n_values = ""56            dose_cum = ""57        else:58            recorded = [float(dose) for dose in doses if dose.strip() != ""]59            n_records = str(len(doses))60            n_values = str(len(recorded))61            dose_cum = format_number(sum(recorded)) if recorded else ""62        records.append(63            {64                "STUDYID": row["STUDYID"],65                "USUBJID": row["USUBJID"],66                "TRT01P": row["ARM"],67                "DOSECUM": dose_cum,68                "NDOSREC": n_records,69                "NDOSVAL": n_values,70            }71        )7273    columns = ["STUDYID", "USUBJID", "TRT01P", "DOSECUM", "NDOSREC", "NDOSVAL"]74    with open(OUTPUT_PATH, "w", newline="", encoding="utf-8") as handle:75        writer = csv.DictWriter(handle, fieldnames=columns)76        writer.writeheader()77        writer.writerows(records)787980if __name__ == "__main__":81    main()

1/* opensas reference solution for the yamaa benchmark adam-adex-uncollected.2   Reads the benchmark inputs and derives the requested output.3   Execute with the existing opensas runtime: opensas /app/output/result.sas. */45data ex;6  length STUDYID DOMAIN USUBJID EXTRT EXDOSU EXSTDTC $1024;7  infile "/app/input/ex.csv" dsd dlm="," firstobs=2 truncover lrecl=32767;8  input STUDYID : $char1024. DOMAIN : $char1024. USUBJID : $char1024. EXSEQ EXTRT : $char1024. EXDOSE EXDOSU : $char1024. EXSTDTC : $char1024.;9  if not missing(EXDOSE) and EXDOSU ne 'mg' then abort return 1;10run;1112data dm;13  length DOMAIN STUDYID USUBJID ARM $1024;14  infile "/app/input/dm.csv" dsd dlm="," firstobs=2 truncover lrecl=32767;15  input DOMAIN : $char1024. STUDYID : $char1024. USUBJID : $char1024. ARM : $char1024.;16run;1718proc sql;19  create table totals as select STUDYID,USUBJID,sum(EXDOSE) as DOSECUM,count(*) as NDOSREC,count(EXDOSE) as NDOSVAL from ex group by STUDYID,USUBJID;20  create table result as select a.STUDYID,a.USUBJID,a.ARM as TRT01P,b.DOSECUM,b.NDOSREC,b.NDOSVAL from dm as a left join totals as b on a.STUDYID=b.STUDYID and a.USUBJID=b.USUBJID;21quit;2223proc export data=result(keep=STUDYID USUBJID TRT01P DOSECUM NDOSREC NDOSVAL) outfile="/app/output/adex.csv" dbms=csv replace; run;