README.md

---
license: cc-by-nc-4.0
task_categories:
- tabular-classification
- tabular-regression
- time-series-forecasting
language:
- en
tags:
- synthetic
- healthcare
- hospital-operations
- operating-room
- or-utilization
- surgical-scheduling
- staffing
- workforce
- nursing-shortage
- equipment
- biomedical
- capacity-planning
- bed-management
- ed-boarding
- aha
- aorn
- nsi
- ecri
- cms-conditions-of-participation
- perioperative
- case-cancellation
- first-case-ontime
- or-turnover
- rn-vacancy
- icu-occupancy
pretty_name: HLT-010 Synthetic Hospital Resource Usage Dataset — OR + Staffing + Equipment + Capacity (Sample Preview)
size_categories:
- 10K<n<100K
---

# HLT-010 — Synthetic Hospital Resource Usage Dataset (Sample Preview)

**A free, schema-identical preview of the full HLT-010 commercial product from [XpertSystems.ai](https://xpertsystems.ai).**

A **fully synthetic** hospital operations dataset combining **operating room schedules**, **staffing/workforce records**, **biomedical equipment utilization**, **daily capacity metrics**, and **facility master data** across mixed facility types (academic / large community / medium community / critical access). Calibrated to AHA Annual Survey 2023, AORN benchmarks, NSI nursing data, ECRI Institute equipment data, and CMS Conditions of Participation.

> ⚠️ **PRIVACY & SYNTHETIC NATURE**
> Every record in this dataset is **100% synthetic**. **No real patient data, no PHI, no real facility identifiers, no real surgeon or staff NPIs.** Population-level distributions match published AHA / AORN / NSI / ECRI benchmark sources but the facilities and operational events are computationally generated.

---

## What's in this sample

| File | Rows | Cols | Description |
|---|---|---|---|
| `facilities.csv` | 3 | 11 | Facility master — type, teaching status, trauma level, bed count, OR suites, PACU bays, region |
| `hospital_resources.csv` | 42 | 41 | Daily capacity + financial + quality KPIs per facility (14 days × 3 facilities) |
| `or_schedule.csv` | ~4,200 | 18 | One row per surgical case — 22 case types, scheduled vs actual timing, cancellations, block ownership |
| `staffing.csv` | ~13,500 | 11 | One row per staff-shift — 12 perioperative roles, OT/float/agency flags, staff-to-patient ratios |
| `equipment.csv` | ~17,500 | 14 | One row per equipment-day — 18 equipment classes, utilization, downtime, maintenance schedule, repair cost |

**Total:** ~3.9 MB across 6 files.

---

## Schema highlights

### `facilities.csv` (11 columns) — facility master

`facility_id`, `facility_type` (academic / large / medium / small), `teaching_status` (Major Teaching / Minor Teaching / Non-Teaching), `trauma_level` (Level I-IV), `bed_count`, `icu_beds`, `or_suite_count`, `pacu_bays`, `state`, `region` (Northeast / Midwest / South / West), `daily_or_capacity`

### `hospital_resources.csv` (41 columns) — daily operational KPIs

**Identity & temporal:** `facility_id`, `census_date`, `day_of_week`, `is_weekend`

**Bed capacity:** `total_beds`, `occupied_beds`, `occupancy_rate`, `icu_beds_x`, `icu_occupied`, `icu_occupancy_rate`, `pacu_bays_x`, `pacu_patients`, `pacu_utilization_rate`

**ED throughput:** `ed_boarding_hours`, `diversion_flag`, `diversion_hours`, `capacity_breach_flag`, `surge_day_flag`

**OR financial & operational:** `or_utilization_rate`, `surgical_cases_scheduled`, `or_revenue_usd`, `or_cost_per_min_usd`, `total_or_minutes`, `contribution_margin_usd`, `block_release_efficiency`

**Quality & safety:** `staffing_adequacy_score`, `operational_efficiency_index`, `surgical_site_infection_flag`, `near_miss_flag`, `consent_timeout_completed`, `equipment_safety_check_flag`

### `or_schedule.csv` (18 columns) — per-case scheduling

`case_id`, `facility_id`, `case_date`, `or_id`, `case_type` (22 types: Orthopedic, Cardiac, General Surgery, Neurosurgery, OB/GYN, Urology, ENT, Plastic Surgery, Vascular, Thoracic, Transplant, Trauma, Ophthalmology, Colorectal, Bariatric, Endoscopy, Interventional Radiology, Gynecologic Oncology, Pediatric Surgery, Spinal, Hand Surgery, Robotic Assisted), `surgeon_id`, `is_emergency`, `scheduled_start_min`, `actual_start_min`, `start_delay_min`, `first_case_ontime_flag`, `scheduled_duration_min`, `actual_duration_min`, `turnover_time_min`, `cancellation_flag`, `cancellation_reason`, `block_owner`, `add_on_flag`

### `staffing.csv` (11 columns) — daily shift records

`shift_id`, `facility_id`, `shift_date`, `staff_id`, `staff_role` (12 roles: Surgeon, Anesthesiologist, CRNA, Scrub Tech, RN Circulator, PA/NP, Resident, Pharmacist, Radiology Tech, Biomedical Tech, Environmental Services, Unit Coordinator), `shift_type` (Day / Evening / Night), `hours_worked`, `overtime_flag`, `float_pool_flag`, `agency_flag`, `staff_to_patient_ratio`

### `equipment.csv` (14 columns) — daily equipment utilization

`asset_id`, `facility_id`, `record_date`, `equipment_class` (18 classes including Anesthesia Machine, Patient Monitor, Infusion Pump, Electrosurgical Unit, Sterilization Autoclave, CT Scanner, MRI Scanner, C-Arm Fluoroscopy, Endoscope Processor, Intraoperative MRI, ECMO Circuit, CRRT Machine, Cardiac Cath Lab Equipment, Defibrillator, Ventilator, Robotic Surgical System, Imaging Workstation, Hybrid OR Imaging), `equipment_age_yrs`, `utilization_rate`, `in_service_hours`, `downtime_hours`, `unplanned_downtime_flag`, `downtime_cause` (Hardware Failure / Software Error / Power Surge / User Error / Calibration Failure / Component Wear / Connectivity Issue / Sensor Malfunction), `last_maintenance_date`, `next_maintenance_due`, `failure_flag`, `repair_cost_usd`

---

## Calibration source story

The full HLT-010 generator anchors all distributions to authoritative hospital operations references:

- **AHA Annual Survey 2023** (American Hospital Association) — OR utilization (78.4%), case cancellations (8.2%), bed occupancy (81.2%), ED boarding (3.2hr), revenue per case (~$18,400)
- **AORN Benchmarks** (Association of periOperative Registered Nurses) — first-case on-time start (82%), OR turnover (28 ± 8 min), surgical tech vacancy (22.8%)
- **NSI Nursing Solutions 2023** — RN vacancy rate (15.6%), turnover patterns
- **ECRI Institute** — Equipment unplanned downtime (~4.2%), age-related failure curves
- **CMS Conditions of Participation** — ICU occupancy target max 85%, staffing-to-patient ratios
- **IHI (Institute for Healthcare Improvement)** — Operational efficiency benchmarks, surge capacity

### Sample-scale validation scorecard

| Metric | Observed | Target | Tolerance | Status | Source |
|---|---|---|---|---|---|
| OR utilization rate | 71.3% | 70% | ±10% | ✅ PASS | AHA 2023 |
| Case cancellation rate | 7.6% | 8% | ±3% | ✅ PASS | AHA 2023 |
| First-case on-time rate | 84.5% | 82% | ±8% | ✅ PASS | AORN Benchmarks |
| OR turnover (min) | 27.4 | 28.0 | ±4.0 | ✅ PASS | AORN |
| Bed occupancy rate | 81.7% | 78% | ±10% | ✅ PASS | AHA 2023 |
| ED boarding hours (mean) | 3.28 | 3.2 | ±1.2 | ✅ PASS | AHA 2023 |
| ICU occupancy (under CMS max) | 80.7% | ≤85% | — | ✅ PASS | CMS CoP |
| Equipment downtime rate | 5.1% | 4.8% | ±1.8% | ✅ PASS | ECRI Institute |
| Case type diversity | 22 | 22 | ±2 | ✅ PASS | AORN surgical taxonomy |
| Staff role diversity | 12 | 12 | — | ✅ PASS | AORN team composition |

**Grade: A+ (100/100) — verified across 6 random seeds (42, 7, 123, 2024, 99, 1).**

---

## Loading examples

### Pandas — explore the operational data

```python
import pandas as pd

facilities = pd.read_csv("facilities.csv")
capacity = pd.read_csv("hospital_resources.csv", parse_dates=["census_date"])
ors = pd.read_csv("or_schedule.csv", parse_dates=["case_date"])
staffing = pd.read_csv("staffing.csv", parse_dates=["shift_date"])
equipment = pd.read_csv("equipment.csv", parse_dates=["record_date"])

# OR utilization by facility type
print(capacity.merge(facilities, on="facility_id")
        .groupby("facility_type")["or_utilization_rate"]
        .agg(["mean", "std", "min", "max"]).round(3))

# Case type mix
print(ors["case_type"].value_counts(normalize=True).head(10).round(3))

# Cancellation reasons
print(ors.loc[ors["cancellation_flag"] == True, "cancellation_reason"]
        .value_counts())
```

### Hugging Face Datasets

```python
from datasets import load_dataset

ds = load_dataset("xpertsystems/hlt010-sample", data_files={
    "facilities":         "facilities.csv",
    "hospital_resources": "hospital_resources.csv",
    "or_schedule":        "or_schedule.csv",
    "staffing":           "staffing.csv",
    "equipment":          "equipment.csv",
})
print(ds)
```

### OR utilization forecasting baseline

```python
import pandas as pd
from sklearn.ensemble import GradientBoostingRegressor
from sklearn.model_selection import train_test_split

cap = pd.read_csv("hospital_resources.csv", parse_dates=["census_date"])
cap["month"] = cap["census_date"].dt.month
cap["dayofweek_num"] = cap["census_date"].dt.dayofweek

features = ["bed_count", "or_suite_count", "is_weekend", "dayofweek_num",
            "month", "occupancy_rate", "icu_occupancy_rate",
            "surgical_cases_scheduled"]
X = cap[features].fillna(0)
y = cap["or_utilization_rate"]
Xtr, Xte, ytr, yte = train_test_split(X, y, test_size=0.3, random_state=42)
m = GradientBoostingRegressor(random_state=42).fit(Xtr, ytr)
print(f"OR utilization R²: {m.score(Xte, yte):.3f}")
```

### Equipment maintenance prediction

```python
import pandas as pd

eq = pd.read_csv("equipment.csv", parse_dates=["record_date"])

# Downtime rate by equipment age
eq["age_bucket"] = pd.cut(eq["equipment_age_yrs"],
                          [0, 3, 6, 10, 15],
                          labels=["0-3yr", "3-6yr", "6-10yr", "10-15yr"])
print(eq.groupby("age_bucket")["unplanned_downtime_flag"].mean().round(3))

# Repair cost distribution
print(eq.loc[eq["repair_cost_usd"] > 0, "repair_cost_usd"].describe())
```

### Staffing analysis

```python
import pandas as pd

staff = pd.read_csv("staffing.csv")

# Overtime rate by role
print(staff.groupby("staff_role")["overtime_flag"].mean().sort_values(ascending=False))

# Agency staff reliance
print(staff.groupby(["facility_id", "staff_role"])["agency_flag"]
        .mean().unstack().round(3))
```

---

## Suggested use cases

- **OR utilization forecasting** — predict next-day OR utilization from facility characteristics + recent operational patterns
- **Surgical case cancellation prediction** — classify cancellation risk to enable proactive intervention
- **Block schedule optimization** — analyze block release efficiency and underutilized blocks
- **Equipment failure prediction** — predict `unplanned_downtime_flag` from age + utilization + maintenance history
- **Maintenance scheduling optimization** — risk-adjusted preventive maintenance interval modeling
- **Staffing-to-acuity matching** — analyze `staff_to_patient_ratio` × `acuity` patterns for nurse scheduling
- **Overtime / agency cost modeling** — predict overtime hours and agency staffing needs
- **Bed capacity surge prediction** — predict `surge_day_flag` and `diversion_flag` from upstream factors
- **ED boarding root cause analysis** — relate `ed_boarding_hours` to ICU occupancy and discharge patterns
- **Quality & safety event modeling** — predict near-miss / SSI / consent timeout events from staffing + acuity
- **Financial contribution margin modeling** — analyze contribution margin drivers across facility types
- **Hospital ML pretraining** — pretrain operational forecasting models before fine-tuning on real EHR/EMR data
- **Operations research education** — perioperative scheduling, queueing theory, capacity planning coursework

---

## Sample vs. full product

| Aspect | This sample | Full HLT-010 product |
|---|---|---|
| Facilities | 3 (mixed) | 50+ (default) up to 500+ |
| Time window | 14 days | 365+ days (multi-year configurable) |
| Facility types | Mixed (3) | Mixed / academic-only / community-only / critical_access |
| Output format | CSV | CSV / Parquet / JSON |
| Schema | identical | identical |
| Calibration | identical | identical |
| License | CC-BY-NC-4.0 | Commercial license |

The full product unlocks:
- **Up to 500+ facilities** for system-wide operations modeling
- **Multi-year longitudinal windows** for trend analysis and intervention impact studies
- **Configurable facility mix** for targeted segmentation (academic-only / community-only / CAH)
- **Parquet output** for production data pipelines
- Commercial use rights

**Contact us for the full product.**

---

## Limitations & honest disclosures

- **Sample is preview-only.** 3 facilities × 14 days × ~35K operational records is enough to demonstrate schema and calibration, but is **not statistically sufficient** for facility-level capacity planning models or season-aware forecasting. Use the full product (50+ facilities × 365 days) for serious work.
- **Sample includes 3 facility types (academic + large + medium), not all 4.** The `critical_access` facility type is not represented at n=3 due to random sampling. The full product reliably covers all 4 types.
- **OR utilization runs slightly below the headline AHA target.** Sample mean is ~70% vs AHA 78.4% pure target. This is partly because mixed facility_mix includes community facilities (which average lower OR utilization) and partly small-N effects at 3 facilities × 14 days. The full product hits the AHA target at scale.
- **Equipment downtime runs slightly elevated (5.1% vs ECRI 4.2%).** The generator's age-based `failure_multiplier` produces realistic but somewhat-higher-than-target downtime for aging assets. Reflects real-world equipment fleet aging — production hospitals with younger fleets see lower rates.
- **PACU utilization clips at 1.0.** The generator caps PACU utilization at 100% rather than allowing over-capacity. At busy academic centers, real PACU congestion exceeds capacity (queue forms) — this is hidden by the cap.
- **Staff IDs are synthetic random integers.** No real NPIs, no real practitioner identifiers. Surgeon IDs are equally synthetic.
- **Equipment IDs are synthetic identifiers**, not real GUDID device IDs.
- **Block-schedule data is daily-aggregated, not minute-level.** The full product can be extended with minute-level block scheduling for highly-detailed OR room optimization.
- **No real ICD-10 / CPT case data joins.** Case types are categorical groupings (Orthopedic, Cardiac, etc.) — the full ICD-10/CPT/MS-DRG joins are in the companion HLT-005 hospital admission dataset.
- **Synthetic, not derived from real hospital operations data.** Distributions match published AHA/AORN/NSI/ECRI references but do NOT reflect any specific real hospital.

---

## Ethical use guidance

This dataset is designed for:
- Hospital operations analytics methodology development
- OR scheduling and capacity planning research
- Equipment maintenance prediction ML
- Nursing workforce analytics
- ED throughput optimization research
- Healthcare AI pretraining for operational forecasting
- Educational use in hospital operations management and operations research

This dataset is **not appropriate for**:
- Making real staffing decisions about real personnel
- Real surgeon performance evaluation
- Real equipment retirement/procurement decisions without validation
- Discriminatory analyses targeting protected demographic groups
- Hospital quality scoring or pay-for-performance modeling without real-data validation

---

## Companion datasets in the Healthcare vertical

- [HLT-001](https://huggingface.co/datasets/xpertsystems/hlt001-sample) — Synthetic Patient Population (5K patients × 79 cols, CDC/NHANES calibrated)
- [HLT-002](https://huggingface.co/datasets/xpertsystems/hlt002-sample) — Synthetic EHR Dataset (4K encounters + FHIR R4 bundles)
- [HLT-003](https://huggingface.co/datasets/xpertsystems/hlt003-sample) — Synthetic Clinical Trial Dataset (3 endpoint types + power sweep)
- [HLT-004](https://huggingface.co/datasets/xpertsystems/hlt004-sample) — Synthetic Disease Progression Dataset (NSCLC + Heart Failure longitudinal)
- [HLT-005](https://huggingface.co/datasets/xpertsystems/hlt005-sample) — Synthetic Hospital Admission Dataset (5K admissions + bed utilization)
- [HLT-006](https://huggingface.co/datasets/xpertsystems/hlt006-sample) — Synthetic Medical Imaging Dataset (1K studies + COCO annotations + reports)
- [HLT-007](https://huggingface.co/datasets/xpertsystems/hlt007-sample) — Synthetic Drug Response Dataset (3K patient-treatments × 25 drug classes + PGx + PK)
- [HLT-008](https://huggingface.co/datasets/xpertsystems/hlt008-sample) — Synthetic Healthcare Claims Dataset (500 members + 30K claims + fraud labels)
- [HLT-009](https://huggingface.co/datasets/xpertsystems/hlt009-sample) — Synthetic Continuous Vital Sign Monitoring Dataset (25 ICU episodes + alarms)
- **HLT-010** — Synthetic Hospital Resource Usage Dataset (you are here)

Use **HLT-001 through HLT-010 together** for the complete healthcare data stack: clinical (population/EHR/trials/progression) + operational (admissions/imaging/pharma/claims/monitoring/**resources**) — 10 datasets covering every major workflow in the modern hospital.

---

## Citation

If you use this dataset, please cite:

```bibtex
@dataset{xpertsystems_hlt010_sample_2026,
  author       = {XpertSystems.ai},
  title        = {HLT-010 Synthetic Hospital Resource Usage Dataset (Sample Preview)},
  year         = 2026,
  publisher    = {Hugging Face},
  url          = {https://huggingface.co/datasets/xpertsystems/hlt010-sample}
}
```

---

## Contact

- **Web:** [https://xpertsystems.ai](https://xpertsystems.ai)
- **Email:** [pradeep@xpertsystems.ai](mailto:pradeep@xpertsystems.ai)
- **Full product catalog:** Cybersecurity, Insurance & Risk, Materials & Energy, Oil & Gas, Healthcare, and more

**Sample License:** CC-BY-NC-4.0 (Creative Commons Attribution-NonCommercial 4.0)
**Full product License:** Commercial — please contact for pricing.