Summarize End-of-Study and End-of-Treatment Disposition

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

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

1# Reference solution for the yamaa benchmark adam-adsl-disposition (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)14ds <- read_csv(15  "/app/input/ds.csv",16  col_types = cols(DSSEQ = col_integer(), DSSTDTC = col_date(), .default = col_character())17)1819# The last dated disposition event per track: latest date, highest sequence20# on a tie. The study track reads every epoch but TREATMENT; the treatment21# track reads only TREATMENT.22last_study <- ds |>23  filter(DSCAT %in% "DISPOSITION EVENT", EPOCH != "TREATMENT", !is.na(DSSTDTC)) |>24  arrange(STUDYID, USUBJID, desc(DSSTDTC), desc(DSSEQ)) |>25  distinct(STUDYID, USUBJID, .keep_all = TRUE) |>26  transmute(STUDYID, USUBJID, EOSDT = DSSTDTC, EOSDECOD = DSDECOD, EOSREAS = DSTERM)2728last_trt <- ds |>29  filter(DSCAT %in% "DISPOSITION EVENT", EPOCH %in% "TREATMENT", !is.na(DSSTDTC)) |>30  arrange(STUDYID, USUBJID, desc(DSSTDTC), desc(DSSEQ)) |>31  distinct(STUDYID, USUBJID, .keep_all = TRUE) |>32  transmute(STUDYID, USUBJID, EOTDT = DSSTDTC, EOTDECOD = DSDECOD, EOTREAS = DSTERM)3334adsl <- dm |>35  left_join(last_study, by = c("STUDYID", "USUBJID")) |>36  left_join(last_trt, by = c("STUDYID", "USUBJID")) |>37  mutate(38    EOSSTT = case_when(39      EOSDECOD == "COMPLETED" ~ "COMPLETED",40      !is.na(EOSDECOD) ~ "DISCONTINUED",41      .default = "ONGOING"42    ),43    DCSREAS = if_else(EOSSTT == "DISCONTINUED", EOSREAS, NA_character_),44    EOTSTT = case_when(45      EOTDECOD == "COMPLETED" ~ "COMPLETED",46      !is.na(EOTDT) ~ "DISCONTINUED",47      .default = "ONGOING"48    ),49    DCTREAS = if_else(EOTSTT == "DISCONTINUED", EOTREAS, NA_character_),50    DCSREASP = if_else(EOTDECOD %in% "OTHER", EOTREAS, NA_character_)51  ) |>52  select(53    STUDYID, USUBJID, EOSDT, EOSDECOD, EOSREAS, EOSSTT, DCSREAS,54    EOTDT, EOTSTT, DCTREAS, DCSREASP55  )5657dir.create("/app/output", showWarnings = FALSE)58write_csv(adsl, "/app/output/adsl.csv", na = "")

1# Reference solution for the yamaa benchmark adam-adsl-disposition (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 1.67import csv89INPUT_DIR = "/app/input"10OUTPUT_DIR = "/app/output"11OUTPUT_FILE = OUTPUT_DIR + "/adsl.csv"121314def read_csv(path):15    with open(path, newline="") as f:16        return list(csv.DictReader(f))171819def last_record(records):20    """Last record: latest DSSTDTC, ties broken by higher DSSEQ."""21    if not records:22        return None23    return max(records, key=lambda r: (r["DSSTDTC"], float(r["DSSEQ"])))242526def main():27    dm = read_csv(INPUT_DIR + "/dm.csv")28    ds = read_csv(INPUT_DIR + "/ds.csv")2930    # Only DISPOSITION EVENT records with a date ever count.31    counted = {}32    for row in ds:33        if row["DSCAT"] != "DISPOSITION EVENT":34            continue35        if not row["DSSTDTC"]:36            continue37        counted.setdefault(row["USUBJID"], []).append(row)3839    fieldnames = [40        "STUDYID", "USUBJID", "EOSDT", "EOSDECOD", "EOSREAS", "EOSSTT",41        "DCSREAS", "EOTDT", "EOTSTT", "DCTREAS", "DCSREASP",42    ]4344    out_rows = []45    for subj in dm:46        usubjid = subj["USUBJID"]47        records = counted.get(usubjid, [])48        study = [r for r in records if r["EPOCH"] != "TREATMENT"]49        treat = [r for r in records if r["EPOCH"] == "TREATMENT"]5051        last_study = last_record(study)52        last_treat = last_record(treat)5354        # End-of-study summary.55        if last_study is None:56            eosdt = eosdecod = eosreas = ""57            eosstt = "ONGOING"58        else:59            eosdt = last_study["DSSTDTC"]60            eosdecod = last_study["DSDECOD"]61            eosreas = last_study["DSTERM"]62            eosstt = "COMPLETED" if eosdecod == "COMPLETED" else "DISCONTINUED"63        dcsreas = eosreas if eosstt == "DISCONTINUED" else ""6465        # End-of-treatment summary.66        if last_treat is None:67            eotdt = ""68            eotstt = "ONGOING"69            treat_term = ""70            treat_decod = ""71        else:72            eotdt = last_treat["DSSTDTC"]73            treat_decod = last_treat["DSDECOD"]74            treat_term = last_treat["DSTERM"]75            eotstt = "COMPLETED" if treat_decod == "COMPLETED" else "DISCONTINUED"76        dctreas = treat_term if eotstt == "DISCONTINUED" else ""77        dcsreasp = treat_term if treat_decod == "OTHER" else ""7879        out_rows.append({80            "STUDYID": subj["STUDYID"],81            "USUBJID": usubjid,82            "EOSDT": eosdt,83            "EOSDECOD": eosdecod,84            "EOSREAS": eosreas,85            "EOSSTT": eosstt,86            "DCSREAS": dcsreas,87            "EOTDT": eotdt,88            "EOTSTT": eotstt,89            "DCTREAS": dctreas,90            "DCSREASP": dcsreasp,91        })9293    with open(OUTPUT_FILE, "w", newline="") as f:94        writer = csv.DictWriter(f, fieldnames=fieldnames)95        writer.writeheader()96        writer.writerows(out_rows)979899if __name__ == "__main__":100    main()

1/* opensas reference solution for the yamaa benchmark adam-adsl-disposition.2   Reads the benchmark inputs and derives the requested output.3   Execute with the existing opensas runtime: opensas /app/output/result.sas. */45data dm;6  length STUDYID USUBJID $1024;7  infile "/app/input/dm.csv" dsd dlm="," firstobs=2 truncover lrecl=32767;8  input STUDYID : $char1024. USUBJID : $char1024.;9run;1011data ds;12  length STUDYID USUBJID DSCAT EPOCH DSDECOD DSTERM DSSTDTC $1024;13  infile "/app/input/ds.csv" dsd dlm="," firstobs=2 truncover lrecl=32767;14  input STUDYID : $char1024. USUBJID : $char1024. DSSEQ DSCAT : $char1024. EPOCH : $char1024. DSDECOD : $char1024. DSTERM : $char1024. DSSTDTC : $char1024.;15run;1617proc sort data=ds(where=(DSCAT='DISPOSITION EVENT' and EPOCH ne 'TREATMENT' and DSSTDTC ne '')) out=ordered_study;18  by STUDYID USUBJID descending DSSTDTC descending DSSEQ;19run;20data lastds; set ordered_study; by STUDYID USUBJID; if first.USUBJID; run;2122proc sort data=ds(where=(DSCAT='DISPOSITION EVENT' and EPOCH='TREATMENT' and DSSTDTC ne '')) out=ordered_trt;23  by STUDYID USUBJID descending DSSTDTC descending DSSEQ;24run;25data lasttrt; set ordered_trt; by STUDYID USUBJID; if first.USUBJID; keep STUDYID USUBJID DSSTDTC DSDECOD DSTERM; run;2627proc sql;28  create table joined as29    select a.*, b.DSSTDTC as EOSDT, b.DSDECOD as EOSDECOD, b.DSTERM as EOSREAS,30           c.DSSTDTC as EOTDT, c.DSDECOD as EOTDECOD, c.DSTERM as EOTREAS31    from dm as a32    left join lastds as b on a.STUDYID=b.STUDYID and a.USUBJID=b.USUBJID33    left join lasttrt as c on a.STUDYID=c.STUDYID and a.USUBJID=c.USUBJID;34quit;35data result;36  length EOSSTT $12 DCSREAS $1024 EOTSTT $12 DCTREAS $1024 DCSREASP $1024;37  set joined;38  if missing(EOSDT) then EOSSTT='ONGOING';39  else if EOSDECOD='COMPLETED' then EOSSTT='COMPLETED';40  else EOSSTT='DISCONTINUED';41  if EOSSTT='DISCONTINUED' then DCSREAS=EOSREAS;42  if missing(EOTDT) then EOTSTT='ONGOING';43  else if EOTDECOD='COMPLETED' then EOTSTT='COMPLETED';44  else EOTSTT='DISCONTINUED';45  if EOTSTT='DISCONTINUED' then DCTREAS=EOTREAS;46  if EOTDECOD='OTHER' then DCSREASP=EOTREAS;47run;4849proc sql;50  create table final as select STUDYID, USUBJID, EOSDT, EOSDECOD, EOSREAS, EOSSTT, DCSREAS, EOTDT, EOTSTT, DCTREAS, DCSREASP from result;51quit;52proc export data=final outfile="/app/output/adsl.csv" dbms=csv replace; run;