Datasetpublic

HipAAsynth/stroke-tribal-icu-cohort-100

Stroke ICU Cohort — Northern Plains Tribal Profile (n=100, synthetic) Summary A synthetic acute-stroke ICU cohort generated against a Northern Plains tribal health profile (rural Indian Health Service service area, American Indian/Alaska Native-majority). It pairs an acute-stroke decision module (NIHSS, tPA eligibility/administration, onset-to-door and door-to-needle timing, modified Rankin outcomes) with a rural, low-connectivity deployment demographic to probe… See the full description on the dataset page: https://huggingface.co/datasets/HipAAsynth/stroke-tribal-icu-cohort-100.

sourceHugging Facecc-by-nc-nd-4.0updated 1mo agoView on Hugging Face

0likes12downloads

Dataset Card

Stroke ICU Cohort — Northern Plains Tribal Profile (n=100, synthetic)

Summary

A synthetic acute-stroke ICU cohort generated against a Northern Plains tribal health profile (rural Indian Health Service service area, American Indian/Alaska Native-majority). It pairs an acute-stroke decision module (NIHSS, tPA eligibility/administration, onset-to-door and door-to-needle timing, modified Rankin outcomes) with a rural, low-connectivity deployment demographic to probe stroke-pathway performance in an under-represented population.

Every record is fully synthetic — generated by a computer algorithm. No real patient data is present.


Rows	100 patients
Columns	37
Clinical module	Stroke (acute ischemic/hemorrhagic decision support)
Population profile	`nd_tribal_region_a`
Engine	HipAAsynth v1.4.0
Schema version	1.1.0
Generation seed	5101 (deterministic)
Generated	2026-08-19

Reproducibility

This cohort is deterministic: regenerating with the same engine version, scenario, and seed reproduces byte-identical output. Each row also carries its own per-patient seed, and the file is fingerprinted below.

—CSV SHA-256: 8927d363c52feec79d740815354cd1f3b82dfcf66142f2523fcc40eef1470d21
—Determinism: verified Run A = Run B (identical SHA-256)

Regenerate with the HipAAsynth engine:

bash

hipaasynth --demo --count 100 --seed 5101 --scenario tribal_stroke --format csv

Schema

37-column structured CSV (cohort.csv):

Column	Description
`patient_id`	Unique synthetic patient identifier
`seed`	Per-patient deterministic generation seed
`age`	Age in years
`sex`	Biological sex
`ethnicity`	Patient ethnicity (profile-weighted)
`height_cm`	Height in centimeters
`weight_kg`	Weight in kilograms
`bmi`	Body mass index
`bmi_category`	BMI classification (lowercase: underweight / normal / overweight / obesityclass1..3)
`conditions`	Semicolon-delimited list of active clinical conditions
`num_visits`	Total encounter visit count
`num_labs`	Total lab order count
`engine_version`	HipAAsynth engine version
`schema_version`	Record schema version
`synthetic`	Always True — no real patient data
`disclaimer`	Per-record synthetic-data disclaimer
`stroke_flag`	Stroke present flag
`stroke_type`	Ischemic / hemorrhagic classification
`nihss_score`	NIH Stroke Scale score
`nihss_category`	NIHSS severity band (mild/moderate/severe)
`tpa_eligible`	tPA thrombolysis eligibility flag
`tpa_administered`	tPA administered flag
`onset_to_door_minutes`	Symptom onset to hospital arrival (minutes)
`door_to_needle_minutes`	Arrival to tPA (minutes)
`sbp_admission`	Systolic BP at admission (mmHg)
`dbp_admission`	Diastolic BP at admission (mmHg)
`glucose_admission`	Glucose at admission (mg/dL)
`atrial_fibrillation`	Atrial fibrillation flag
`prior_stroke`	Prior stroke flag
`prior_ich`	Prior intracranial hemorrhage flag
`recent_surgery`	Recent surgery flag
`anticoagulant_use`	Anticoagulant use flag
`mrs_discharge`	Modified Rankin Scale at discharge (0-6)
`mrs_90day`	Modified Rankin Scale at 90 days (0-6)
`rural_presentation`	Rural presentation flag
`region_profile`	Population profile used for generation
`stroke_observation_version`	Stroke observation generator version

License


Engine	AGPL-3.0-or-later (a separate commercial license is also available)
Data	CC BY-NC-ND 4.0

This dataset is provided for non-commercial research, transparency, and audit purposes under the CC BY-NC-ND 4.0 license. Commercial use — including model training for profit or benchmarking by commercial entities — requires a separate license. For larger cohorts or commercial inquiries, contact hipaasynth@gmail.com · https://hipaasynth.com