ziadatalabs/FreeSyntheticEHRPatientJourneys

Free Synthetic EHR Patient Journeys

Fully synthetic electronic-health-record-style dataset: 2 million patients and their linked care journeys across five relational tables — patients, encounters, diagnoses, labs, and prescriptions. Built to model realistic longitudinal structure (a patient's chronic condition drives their encounter frequency, diagnosis pattern, lab trends, and prescriptions over time) without containing any real patient data, real clinical records, or real medical guidance. Structural and statistical only — not for clinical decision-making.

Schema

ehr_patients (2,000,000 rows)

Column	Type	Description
patient_id	string	Unique patient identifier
birth_year	int32	Year of birth
sex	string	M / F
region	string	One of 6 US-style regions
enrollment_date	string (date)	Date patient enrolled
chronic_condition	string	none / diabetes / hypertension / asthma / multiple
treatment_adherence	string	n/a / good / moderate / poor (drives lab trend direction)
risk_score	float	Composite risk score (age + condition + noise)

ehr_encounters (9,005,747 rows)

Column	Type	Description
encounter_id	string	Unique encounter identifier
patient_id	string	FK → ehr_patients
encounter_date	string (date)	Date of encounter
encounter_type	string	outpatient / telehealth / ER / inpatient
department	string	Department seen
provider_id	string	Provider identifier (pool of 5,000)
reason_code	string	Encounter reason

ehr_diagnoses (13,509,656 rows)

Column	Type	Description
diagnosis_id	string	Unique diagnosis identifier
encounter_id	string	FK → ehr_encounters
patient_id	string	FK → ehr_patients
diagnosis_code	string	ICD-10-style code
diagnosis_desc	string	Diagnosis description
is_chronic	bool	Whether this ties to the patient's chronic condition
sequence_number	int32	Order within encounter

ehr_labs (15,903,125 rows)

Column	Type	Description
lab_id	string	Unique lab result identifier
encounter_id	string	FK → ehr_encounters
patient_id	string	FK → ehr_patients
test_code	string	Lab test code
test_name	string	Lab test name
result_value	float	Result value
unit	string	Unit of measure
reference_range	string	Normal reference range
is_abnormal	bool	Outside reference range
test_date	string (date)	Date of test

ehr_prescriptions (3,721,239 rows)

Column	Type	Description
prescription_id	string	Unique prescription identifier
encounter_id	string	FK → ehr_encounters
patient_id	string	FK → ehr_patients
drug_class	string	Pharmacological class (synthetic, not a specific drug/dose recommendation)
dosage	string	low / standard / high (bucketed, not clinical guidance)
frequency	string	Dosing frequency bucket
start_date	string (date)	Prescription start date
duration_days	int32	Duration in days
is_chronic_med	bool	Tied to an ongoing chronic condition

Format

Snappy-compressed Parquet, one file per table.

Quick start

import pandas as pd
patients = pd.read_parquet("ehr_patients.parquet")
encounters = pd.read_parquet("ehr_encounters.parquet")

from datasets import load_dataset
ds = load_dataset("ziadatalabs/FreeSyntheticEHRPatientJourneys")

import duckdb
duckdb.sql("SELECT * FROM 'ehr_patients.parquet' LIMIT 10").show()

Notes

• Fully synthetic — no real patients, no real clinical records.
• Chronic-condition patients show consistent, correlated trajectories across tables (more encounters, condition-linked diagnoses/labs/prescriptions, and lab values that trend with a treatment-adherence flag).
• Structural/statistical only — not intended for clinical use or medical decision-making.

License & Usage

CC BY-NC 4.0. Free for non-commercial use.

Want more free datasets? Hit the ❤️ and follow. And we take requests — tell us what synthetic data you need, and we'll build it.

---

Zia Data Labs — zia.data.team@protonmail.com