---
license: cc-by-4.0
task_categories:
  - tabular-classification
  - tabular-regression
  - time-series-forecasting
language:
  - en
tags:
  - synthetic
  - healthcare
  - behavioral-health
  - patients
  - clinical-operations
  - electronic-health-records
  - EHR
  - EMR
  - telehealth
pretty_name: Solstice Behavioral Health Pack
size_categories:
  - 10K<n<100K
configs:
  - config_name: assessments
    data_files:
      - split: train
        path: data/assessments/train.csv
  - config_name: bed_capacity_census
    data_files:
      - split: train
        path: data/bed_capacity_census/train.csv
  - config_name: dashboard_suggestions
    data_files:
      - split: train
        path: data/dashboard_suggestions/train.csv
  - config_name: de_escalation_interventions
    data_files:
      - split: train
        path: data/de_escalation_interventions/train.csv
  - config_name: discharge_plans
    data_files:
      - split: train
        path: data/discharge_plans/train.csv
  - config_name: episodes_of_care
    data_files:
      - split: train
        path: data/episodes_of_care/train.csv
  - config_name: facilities
    data_files:
      - split: train
        path: data/facilities/train.csv
  - config_name: follow_up_events
    data_files:
      - split: train
        path: data/follow_up_events/train.csv
  - config_name: medication_management
    data_files:
      - split: train
        path: data/medication_management/train.csv
  - config_name: metric_definitions
    data_files:
      - split: train
        path: data/metric_definitions/train.csv
  - config_name: patients_synthetic
    data_files:
      - split: train
        path: data/patients_synthetic/train.csv
  - config_name: payer_authorization
    data_files:
      - split: train
        path: data/payer_authorization/train.csv
  - config_name: programs
    data_files:
      - split: train
        path: data/programs/train.csv
  - config_name: quality_metrics_daily
    data_files:
      - split: train
        path: data/quality_metrics_daily/train.csv
  - config_name: referrals
    data_files:
      - split: train
        path: data/referrals/train.csv
  - config_name: restraint_seclusion_events
    data_files:
      - split: train
        path: data/restraint_seclusion_events/train.csv
  - config_name: safety_events
    data_files:
      - split: train
        path: data/safety_events/train.csv
  - config_name: staffing_shifts
    data_files:
      - split: train
        path: data/staffing_shifts/train.csv
---

# Solstice Behavioral Health Pack (Sample)

**A comprehensive synthetic dataset for behavioral health operations.** This dataset simulates clinical workflows, patient trajectories, and facility metrics for inpatient and outpatient behavioral health environments.

Built by [Solstice AI Studio](https://www.solsticestudio.ai/datasets) as a free sample of a larger commercial pack. 100% synthetic — no real patient, provider, or facility data.

## What's in the box

This dataset consists of 18 interconnected tables covering:
- **Patient Dimensions:** `patients_synthetic`, `facilities`, `programs`
- **Clinical Fact Tables:** `episodes_of_care`, `assessments`, `medication_management`
- **Operational Metrics:** `bed_capacity_census`, `staffing_shifts`, `quality_metrics_daily`
- **Safety & Quality:** `safety_events`, `restraint_seclusion_events`, `de_escalation_interventions`
- **Administrative:** `referrals`, `payer_authorization`, `discharge_plans`, `follow_up_events`

## Use Cases
- **Clinical Predictive Modeling:** Train models to predict readmission risk or length of stay.
- **Operational Efficiency:** Benchmarking staffing rollups vs. bed capacity.
- **Safety Analytics:** Identifying correlations between acuity levels and safety events.
- **RAG & Agent Benchmarking:** Testing autonomous agents on complex healthcare schemas.

## Data Provenance
Generated using Solstice’s PhantasOS / SIMA simulation engine. The simulation models medically consistent state transitions for behavioral health conditions (e.g., Depression, SUD, Sepsis) and administrative workflows matching standard EHR patterns.

## Responsible Use Statement
Standard packs do not contain PHI or real patient records. They are intended for research, testing, prototyping, and model development, not clinical decision-making.

## Get the Full Pack
Scale this dataset to 1M+ episodes, custom facilities, and multi-year timelines.
[www.solsticestudio.ai/datasets](https://www.solsticestudio.ai/datasets)