Getting started

Access CDISC ADaM and SDTM test datasets in Python

Datasets are module attributes, lazy-loaded from parquet.

Collections

Module	Standard	Source	Datasets
pharmadata.pharmaverseadam	ADaM	pharmaverseadam	30
pharmadata.pharmaversesdtm	SDTM	pharmaversesdtm	64
pharmadata.cdiscpilotadam	ADaM	CDISC pilot	12
pharmadata.cdiscpilotsdtm	SDTM	CDISC pilot	31

Install

pip install pharmadata

Usage

import polars as pl
from pharmadata import pharmaverseadam as adam, pharmaversesdtm as sdtm

adam.list_datasets()
adam.meta_tables()
adam.meta_columns()
adam.meta_specs()
adam_dict = adam.to_dict()

adsl = adam.adsl
adsl_meta = adam.adsl_meta
adsl_meta.column_names
adsl_meta.column_labels
adsl_meta.column_names_to_labels
adsl_meta.number_rows
adsl_meta.number_columns
adsl_meta.file_label
adsl_meta.file_encoding
adsl.head()

shape: (5, 55)

STUDYID	USUBJID	SUBJID	SITEID	COUNTRY	RFSTDTC	RFENDTC	RFXSTDTC	RFXENDTC	RFPENDTC	SCRFDT	FRVDT	DTHDTC	DTHADY	DTHFL	LDDTHELD	LDDTHGR1	DTH30FL	DTHA30FL	DTHDOM	DTHB30FL	REGION1	DMDTC	DMDY	AGE	AGEU	AGEGR1	SEX	RACE	RACEGR1	ETHNIC	SAFFL	ARM	ARMCD	ACTARM	ACTARMCD	TRT01P	TRT01A	TRTSDT	TRTSDTM	TRTSTMF	TRTEDT	TRTEDTM	TRTETMF	EOSSTT	EOSDT	RFICDTC	RANDDT	LSTALVDT	TRTDURD	DTHDT	DTHDTF	DTHCAUS	DTHCGR1	BRTHDTC
str	str	str	str	str	str	str	str	str	str	date	date	str	f64	str	f64	str	str	str	str	str	str	str	f64	f64	str	str	str	str	str	str	str	str	str	str	str	str	str	date	datetime[μs, UTC]	str	date	datetime[μs, UTC]	str	str	date	str	date	date	f64	date	str	str	str	str
"CDISCPILOT01"	"01-701-1015"	"1015"	"701"	"USA"	"2014-01-02"	"2014-07-02"	"2014-01-02"	"2014-07-02"	"2014-07-02T11:45"	null	null	null	null	null	null	null	null	null	null	null	"NA"	"2013-12-26"	-7.0	63.0	"YEARS"	"18-64"	"F"	"WHITE"	"White"	"HISPANIC OR LATINO"	"Y"	"Placebo"	"Pbo"	"Placebo"	"Pbo"	"Placebo"	"Placebo"	2014-01-02	2014-01-02 00:00:00 UTC	"H"	2014-07-02	2014-07-02 23:59:59 UTC	"H"	"COMPLETED"	2014-07-02	null	2014-01-02	2014-07-02	182.0	null	null	null	null	"1950-12-26"
"CDISCPILOT01"	"01-701-1023"	"1023"	"701"	"USA"	"2012-08-05"	"2012-09-02"	"2012-08-05"	"2012-09-01"	"2013-02-18"	null	2013-02-18	null	null	null	null	null	null	null	null	null	"NA"	"2012-07-22"	-14.0	64.0	"YEARS"	"18-64"	"M"	"WHITE"	"White"	"HISPANIC OR LATINO"	"Y"	"Placebo"	"Pbo"	"Placebo"	"Pbo"	"Placebo"	"Placebo"	2012-08-05	2012-08-05 00:00:00 UTC	"H"	2012-09-01	2012-09-01 23:59:59 UTC	"H"	"DISCONTINUED"	2012-09-02	null	2012-08-05	2012-09-02	28.0	null	null	null	null	"1948-07-22"
"CDISCPILOT01"	"01-701-1028"	"1028"	"701"	"USA"	"2013-07-19"	"2014-01-14"	"2013-07-19"	"2014-01-14"	"2014-01-14T11:10"	null	null	null	null	null	null	null	null	null	null	null	"NA"	"2013-07-11"	-8.0	71.0	"YEARS"	">64"	"M"	"WHITE"	"White"	"NOT HISPANIC OR LATINO"	"Y"	"Xanomeline High Dose"	"Xan_Hi"	"Xanomeline High Dose"	"Xan_Hi"	"Xanomeline High Dose"	"Xanomeline High Dose"	2013-07-19	2013-07-19 00:00:00 UTC	"H"	2014-01-14	2014-01-14 23:59:59 UTC	"H"	"COMPLETED"	2014-01-14	null	2013-07-19	2014-01-14	180.0	null	null	null	null	"1942-07-11"
"CDISCPILOT01"	"01-701-1033"	"1033"	"701"	"USA"	"2014-03-18"	"2014-04-14"	"2014-03-18"	"2014-03-31"	"2014-09-15"	null	2014-09-15	null	null	null	null	null	null	null	null	null	"NA"	"2014-03-10"	-8.0	74.0	"YEARS"	">64"	"M"	"WHITE"	"White"	"NOT HISPANIC OR LATINO"	"Y"	"Xanomeline Low Dose"	"Xan_Lo"	"Xanomeline Low Dose"	"Xan_Lo"	"Xanomeline Low Dose"	"Xanomeline Low Dose"	2014-03-18	2014-03-18 00:00:00 UTC	"H"	2014-03-31	2014-03-31 23:59:59 UTC	"H"	"DISCONTINUED"	2014-04-14	null	2014-03-18	2014-04-14	14.0	null	null	null	null	"1940-03-10"
"CDISCPILOT01"	"01-701-1034"	"1034"	"701"	"USA"	"2014-07-01"	"2014-12-30"	"2014-07-01"	"2014-12-30"	"2014-12-30T09:50"	null	null	null	null	null	null	null	null	null	null	null	"NA"	"2014-06-24"	-7.0	77.0	"YEARS"	">64"	"F"	"WHITE"	"White"	"NOT HISPANIC OR LATINO"	"Y"	"Xanomeline High Dose"	"Xan_Hi"	"Xanomeline High Dose"	"Xan_Hi"	"Xanomeline High Dose"	"Xanomeline High Dose"	2014-07-01	2014-07-01 00:00:00 UTC	"H"	2014-12-30	2014-12-30 23:59:59 UTC	"H"	"COMPLETED"	2014-12-30	null	2014-07-01	2014-12-30	183.0	null	null	null	null	"1937-06-24"

Format

from pharmadata import set_output_pandas, set_output_polars, output_format
set_output_pandas()
dm = sdtm.dm
print(type(dm))
set_output_polars()
dm = sdtm.dm
print(type(dm))

<class 'pandas.DataFrame'>
<class 'polars.dataframe.frame.DataFrame'>

set_output_pandas() / set_output_polars() switch globally; output_format() is a context manager.

with output_format("pandas"):
    adsl_pd = adam.adsl

print(type(adsl_pd))

<class 'pandas.DataFrame'>

Validate by pointblank

import pointblank as pb
# pb.validate_adam(adsl, "ADSL")
pb.validate_adam_structure(adsl, "ADSL")

{'dataset': 'ADSL',
 'dataset_class': 'ADSL',
 'valid': True,
 'missing_required': [],
 'missing_conditional': [],
 'unknown_variables': [],
 'population_flags_found': ['SAFFL'],
 'issues': []}

Polars Examples

adsl.select(["STUDYID", "USUBJID", "AGE", "SEX", "TRT01A"]).head()

shape: (5, 5)

STUDYID	USUBJID	AGE	SEX	TRT01A
str	str	f64	str	str
"CDISCPILOT01"	"01-701-1015"	63.0	"F"	"Placebo"
"CDISCPILOT01"	"01-701-1023"	64.0	"M"	"Placebo"
"CDISCPILOT01"	"01-701-1028"	71.0	"M"	"Xanomeline High Dose"
"CDISCPILOT01"	"01-701-1033"	74.0	"M"	"Xanomeline Low Dose"
"CDISCPILOT01"	"01-701-1034"	77.0	"F"	"Xanomeline High Dose"

adsl.filter((pl.col("AGE") >= 75) & (pl.col("SEX") == "F")).select(
    ["USUBJID", "AGE", "SEX", "TRT01A"]
).head()

shape: (5, 4)

USUBJID	AGE	SEX	TRT01A
str	f64	str	str
"01-701-1034"	77.0	"F"	"Xanomeline High Dose"
"01-701-1047"	85.0	"F"	"Placebo"
"01-701-1111"	81.0	"F"	"Xanomeline Low Dose"
"01-701-1133"	81.0	"F"	"Xanomeline High Dose"
"01-701-1146"	75.0	"F"	"Xanomeline High Dose"

adsl.filter(pl.col("TRT01A") != "Screen Failure").group_by("TRT01A").agg(
    pl.col("AGE").mean().round(1).alias("mean_age"),
    pl.len().alias("n"),
).sort("TRT01A")

shape: (3, 3)

TRT01A	mean_age	n
str	f64	u32
"Placebo"	75.2	86
"Xanomeline High Dose"	73.8	72
"Xanomeline Low Dose"	76.0	96

Read parquet directly

• Dataset label in schema metadata
• Column labels in field metadata.

Both are bytes, decode to str

from importlib import resources

import pyarrow.parquet as pq

path = resources.files("pharmadata") / "_data" / "pharmaverseadam" / "adsl.parquet"
schema = pq.ParquetFile(path).schema_arrow

print((schema.metadata or {}).get(b"label", b"").decode())
print({
    name: (schema.field(name).metadata or {}).get(b"label", b"").decode()
    for name in schema.names[:3]
})

Subject Level Analysis
{'STUDYID': 'Study Identifier', 'USUBJID': 'Unique Subject Identifier', 'SUBJID': 'Subject Identifier for the Study'}

pl.from_arrow(pq.read_table(path, columns=schema.names[:6])).head(5)

shape: (5, 6)

STUDYID	USUBJID	SUBJID	SITEID	COUNTRY	RFSTDTC
str	str	str	str	str	str
"CDISCPILOT01"	"01-701-1015"	"1015"	"701"	"USA"	"2014-01-02"
"CDISCPILOT01"	"01-701-1023"	"1023"	"701"	"USA"	"2012-08-05"
"CDISCPILOT01"	"01-701-1028"	"1028"	"701"	"USA"	"2013-07-19"
"CDISCPILOT01"	"01-701-1033"	"1033"	"701"	"USA"	"2014-03-18"
"CDISCPILOT01"	"01-701-1034"	"1034"	"701"	"USA"	"2014-07-01"