Skip to content

CSA Loom — the Microsoft Fabric experience for Azure tenants where Fabric isn't yet available: lakehouses, warehouses, notebooks, semantic models, Activator rules, Data Agents, across Commercial, GCC, GCC-High, and DoD IL5

Synthetic clinical encounters dataset (CUI-safe / no-PHI)¶

A fabricated healthcare encounter set for HIPAA-regulated and pharma workshops (Commercial CoE). Used as a Day-3 alternate workload.

No PHI — synthetic by construction

Every patient, encounter, provider, and diagnosis is machine-generated. No real patient data, MRNs, or PHI are present. Patient IDs are obviously synthetic (PT-…) and do not map to any real individual. This dataset is safe under HIPAA because it contains no protected health information — it is fabricated for instruction only.

Files¶

File	Purpose	Rows (workshop size)
`encounters.csv`	Encounter fact (Bronze source)	~30,000
`patients.csv`	Synthetic patient dimension	~5,000
`providers.csv`	Provider dimension	~150

Schema — `encounters`¶

Column	Type	Notes
`encounter_id`	string (UUID)	Synthetic surrogate key
`patient_id`	string	FK → `patients.patient_id` (e.g., `PT-024189`)
`provider_id`	string	FK → `providers.provider_id`
`encounter_date`	date	Synthetic, spread across ~1 year
`encounter_type`	string	`inpatient` \| `outpatient` \| `emergency` \| `telehealth`
`primary_dx_code`	string	Synthetic ICD-10-like code (fabricated, not real coding)
`length_of_stay_days`	int	0 for outpatient/telehealth
`readmission_30d`	boolean	Synthetic outcome label for the analytics lab

Schema — `patients`¶

Column	Type	Notes
`patient_id`	string	PK (synthetic; not an MRN)
`age_band`	string	`0-17` \| `18-39` \| `40-64` \| `65+` (banded — never a DOB)
`sex`	string	Generic category
`region`	string	Generic label, never an address

Schema — `providers`¶

Column	Type	Notes
`provider_id`	string	PK (synthetic; not an NPI)
`specialty`	string	e.g., `cardiology`, `family_medicine`
`facility`	string	Fabricated facility name

Sample rows — `encounters`¶

encounter_id,patient_id,provider_id,encounter_date,encounter_type,primary_dx_code,length_of_stay_days,readmission_30d
9a1b...,PT-024189,PRV-0042,2026-02-11,inpatient,SYN-I50,4,true
8b2c...,PT-024189,PRV-0042,2026-03-02,outpatient,SYN-I50,0,false
7c3d...,PT-031007,PRV-0118,2026-02-15,emergency,SYN-J18,1,false

De-identification by design

Direct identifiers (names, MRNs, DOBs, addresses) are absent — not masked, absent. Age is banded; diagnosis codes are synthetic (SYN-…) prefixes, not real ICD-10. This keeps the dataset outside PHI scope entirely.

Lab use¶

Day 3 (Transform, healthcare alternate): Bronze → Silver conforms encounter types; Silver → Gold builds gold.readmission_rate_by_specialty (a common quality metric) using the synthetic readmission_30d label.
Day 4 (Data Agent): ground an agent on the Gold quality tables for questions like "which specialty had the highest 30-day readmission rate".