README.md

---
license: cc-by-nc-4.0
task_categories:
- tabular-classification
- tabular-regression
language:
- en
tags:
- synthetic
- healthcare
- rare-disease
- orphan-drug
- orpha
- omim
- clinvar
- acmg
- hpo
- gene-therapy
- enzyme-replacement-therapy
- pharmacogenomics
- clinical-trial-design
- trial-eligibility
- screening-funnel
- adaptive-randomization
- protocol-deviation
- nord
- diagnostic-odyssey
- inheritance-patterns
- ich-gcp
- ctcae
- gaucher
- sma
- cystic-fibrosis
- dmd
- fabry
- pompe
- huntington
- als
- alzheimer
- leber-congenital-amaurosis
pretty_name: HLT-011 Synthetic Rare Disease Cohort + Clinical Trial Eligibility Engine (Sample Preview)
size_categories:
- 1K<n<10K
---

# HLT-011 — Synthetic Rare Disease Dataset + Clinical Trial Eligibility Engine (Sample Preview)

**A free, schema-identical preview of the full HLT-011 commercial product from [XpertSystems.ai](https://xpertsystems.ai).**

A **fully synthetic** rare disease dataset combining patient-level clinical phenotypes (50 ORPHA/OMIM-indexed diseases, ACMG variant classification, HPO term coding, genetic inheritance patterns, NORD-calibrated diagnostic odyssey, treatment history, biomarkers, QoL, insurance access) with a **15-trial clinical eligibility engine** that scores each patient against disease-specific I/E criteria, simulates screening funnels, performs adaptive randomization, and tracks primary endpoints with protocol deviation detection.

> ⚠️ **PRIVACY & SYNTHETIC NATURE**
> Every record in this dataset is **100% synthetic**. **No real patient data, no PHI, no real gene variants, no real clinical trial enrollments.** Population-level distributions match published NORD / ORPHA / ClinVar / ACMG / HPO / ICH GCP / Genetic Alliance benchmarks but the patients and trials are computationally generated.

---

## What's in this sample

### Rare disease cohort (1,500 patients × 107 columns)

`rare_disease_cohort.csv` — One row per patient with **107 columns** spanning:

- **Disease & genetics:** `disease_name`, `disease_orpha_code`, `omim_id`, `disease_group` (16 groups), `prevalence_per_100k`, `prevalence_class`, `ultra_rare_flag`, `gene_symbol`, `variant_hgvs_cdna`, `variant_hgvs_protein`, `variant_type`, `zygosity`, `inheritance_pattern` (AD/AR/XL/XLR/XLD/De novo), `acmg_classification` (P/LP/VUS/LB/B), `clinvar_id`, `penetrance`, `de_novo_flag`, `modifier_gene_flag`, `modifier_gene`
- **Phenotype (HPO):** 25 HPO term slots (`hpo_term_1` through `hpo_term_25`), `n_hpo_terms`, `phenotype_onset`, `phenotype_progression`, `phenotype_severity_score`
- **Diagnostic odyssey:** `symptom_onset_age`, `first_specialist_age`, `genetic_test_age`, `diagnosis_age`, `diagnostic_delay_years`, `n_misdiagnoses`, `test_type`, `test_year`, `result_delay_weeks`, `vus_reclassification_flag`, `nbs_result` (newborn screening), `family_history_flag`, `n_carrier_relatives`
- **Clinical status:** `functional_status_score`, `ambulatory_loss_age`, `cognitive_decline_flag`, `hospitalization_annual_rate`, `icu_admission_flag`, `avg_hosp_los_days`
- **Treatment:** `treatment_category` (ERT/SRT/gene_therapy/symptom_management/none), `drug_name_generic`, `investigational_flag`, `treatment_start_age`, `response_status`, `discontinuation_reason`, `ae_ctcae_grade`, `conmed_count`
- **Biomarkers:** `enzyme_activity_baseline`, `enzyme_activity_6mo/12mo/24mo`, `substrate_level_baseline/12mo`, `protein_biomarker_level`, `metabolite_panel_score`, `imaging_biomarker_value`, `biomarker_trajectory_flag`
- **Trial readiness:** `eligibility_score`, `enrolled_flag`, `trial_template_id`, `trial_arm`, `primary_endpoint_delta_pct`, `protocol_deviation_flag`
- **Survival & QoL:** `survival_probability_5yr`, `qol_eq5d_baseline/12mo`, `prom_score_baseline/12mo`, `caregiver_burden_zbi`
- **Access:** `registry_enrolled_flag`, `patient_advocacy_flag`, `insurance_coverage_flag` (Full/Partial/Denied/Unknown), `distance_specialty_center_miles`, `care_complexity_score`

### Clinical trial eligibility engine outputs

| File | Rows | Description |
|---|---|---|
| `trial_templates.json` | 15 trials | Full protocol designs — I/E criteria, arms, primary/secondary endpoints |
| `trial_eligibility.csv` | ~7,500 | Per-patient × per-trial eligibility scores (10-criterion weighted composite) |
| `trial_enrollment.csv` | ~467 | Enrolled patients with arm assignment (adaptive randomization) |
| `trial_endpoints.csv` | ~5,800 | Longitudinal endpoint assessments — responder flags, protocol deviations |
| `trial_screening_funnel.csv` | 15 | Per-trial: screened → eligible → enrolled → completed funnel |
| `trial_enrichment.csv` | 15 | Per-trial top eligibility blocker + enrichment ratio |

### Documentation

| File | Description |
|---|---|
| `cohort_benchmark_summary.json` | Generator's internal 15-check Grade A+ verification |
| `trial_engine_summary.json` | Engine diagnostics (enrollment stats, responder rates, deviation rates) |
| `validation_scorecard.json` | Wrapper-authored 10-metric scorecard with named sources |

**Total:** ~1.6 MB across 10 files.

---

## Coverage — 50 rare diseases across 16 groups, 15 clinical trials

### Disease catalog (50 ORPHA/OMIM-indexed)

**Metabolic (Lysosomal Storage Disorders):** Gaucher Disease Type 1, Fabry Disease, Pompe Disease, MPS I (Hurler), MPS II (Hunter), Niemann-Pick Type C, Phenylketonuria, Wilson Disease, Tyrosinemia Type I, Homocystinuria, Maple Syrup Urine Disease

**Neurological:** Spinal Muscular Atrophy, Huntington Disease, ALS, Friedreich Ataxia, Rett Syndrome, Angelman Syndrome, Tuberous Sclerosis, Ataxia-Telangiectasia

**Muscular:** Duchenne Muscular Dystrophy, Becker Muscular Dystrophy, Myotonic Dystrophy, Limb-Girdle Muscular Dystrophy

**Pulmonary:** Cystic Fibrosis, Pulmonary Arterial Hypertension, Alpha-1 Antitrypsin Deficiency

**Cardiac:** Marfan Syndrome, Hypertrophic Cardiomyopathy, Long QT Syndrome, Brugada Syndrome

**Hematological:** Sickle Cell Disease, Beta-Thalassemia, Hemophilia A, Hemophilia B, Hereditary Angioedema, Diamond-Blackfan Anemia, Fanconi Anemia

**Endocrine:** Congenital Adrenal Hyperplasia, X-linked Hypophosphatemia

**Renal:** Polycystic Kidney Disease, Alport Syndrome, Cystinosis

**Skeletal:** Osteogenesis Imperfecta, Achondroplasia, Hypophosphatasia

**Ophthalmic:** Leber Congenital Amaurosis, Retinitis Pigmentosa, Stargardt Disease

**Oncological:** Neurofibromatosis Type 1, Von Hippel-Lindau, Li-Fraumeni Syndrome

**Other:** Various chromosomal, immunological, dermatological, hepatic, and auditory rare disorders.

### 15 Clinical trial templates (Phase 2 & 3)

| Trial ID | Disease | Phase | N target | Intervention |
|---|---|---|---|---|
| TRIAL-GD-001 | Gaucher Disease Type 1 | Phase 3 | 120 | Imiglucerase ERT vs Eliglustat SRT |
| TRIAL-SMA-001 | Spinal Muscular Atrophy | Phase 3 | 80 | Nusinersen ASO vs Onasemnogene gene therapy |
| TRIAL-CF-001 | Cystic Fibrosis | Phase 3 | 200 | Triple CFTR modulator combination |
| TRIAL-DMD-001 | Duchenne Muscular Dystrophy | Phase 3 | 60 | Exon-skipping ASO |
| TRIAL-HD-001 | Huntington Disease | Phase 2 | 100 | HTT-lowering ASO |
| TRIAL-FAB-001 | Fabry Disease | Phase 3 | 90 | Migalastat vs Agalsidase ERT |
| TRIAL-MFS-001 | Marfan Syndrome | Phase 3 | 150 | Losartan + Atenolol |
| TRIAL-HEM-001 | Hemophilia A | Phase 3 | 70 | Emicizumab vs Factor VIII |
| TRIAL-PKD-001 | Polycystic Kidney Disease | Phase 3 | 180 | Tolvaptan |
| TRIAL-OI-001 | Osteogenesis Imperfecta | Phase 3 | 100 | Setrusumab anti-sclerostin |
| TRIAL-TS-001 | Tuberous Sclerosis | Phase 3 | 80 | Sirolimus mTOR inhibitor |
| TRIAL-SCD-001 | Sickle Cell Disease | Phase 3 | 75 | Voxelotor / Crizanlizumab |
| TRIAL-NPC-001 | Niemann-Pick Type C | Phase 3 | 50 | Arimoclomol vs Placebo |
| TRIAL-LCA-001 | Leber Congenital Amaurosis | Phase 3 | 40 | Voretigene neparvovec gene therapy |
| TRIAL-CAH-001 | Congenital Adrenal Hyperplasia | Phase 3 | 110 | Crinecerfont (CRF1 receptor antagonist) |

### Eligibility scoring engine (10-criterion weighted composite)

Each patient gets a per-trial eligibility score from these weighted criteria:

1. **Age window** (must fall within trial age criteria)
2. **Gene match** (must have the required gene)
3. **ACMG classification** (P/LP preferred; VUS gets partial credit)
4. **Functional status** (≤ trial's max functional status)
5. **Safety/AE grade** (current AE grade < cutoff)
6. **Prior treatment** (per trial's washout requirement)
7. **Insurance access** (Full > Partial > Denied)
8. **Geographic access** (distance to specialty center)
9. **Disease-specific biomarker** (e.g., enzyme activity)
10. **Composite eligibility threshold** per trial

---

## Calibration source story

The full HLT-011 generator anchors all distributions to authoritative rare disease references:

- **NORD (National Organization for Rare Disorders)** — Diagnostic Odyssey Survey 2020; 5.6 yr delay, 2.8 misdiagnoses
- **Orphanet (ORPHA codes)** — Rare disease prevalence registry
- **OMIM (Online Mendelian Inheritance in Man)** — Genetic disease catalog
- **ACMG Standards and Guidelines (Richards et al. 2015)** — Variant classification framework
- **ClinVar** — Variant interpretation database; clinical cohort ACMG distributions
- **HPO (Human Phenotype Ontology)** — Phenotype term coding
- **ICH GCP E6(R2)** — Good Clinical Practice protocol deviation standards
- **CTCAE v5.0 (NCI)** — Adverse event grading
- **Genetic Alliance** — Inheritance pattern distribution in rare disease
- **Augustine et al. (2023) Nat Rev Drug Discov** — Rare disease trial responder rates
- **FDA Orphan Drug Designation** database — Phase 2/3 trial design conventions

### Sample-scale validation scorecard

| Metric | Observed | Target | Tolerance | Status | Source |
|---|---|---|---|---|---|
| Disease diversity count | 48 | ≥45 of 50 | ±5 | ✅ PASS | Orphanet catalog |
| Disease group count | 16 | ≥14 of 16 | ±2 | ✅ PASS | Orphanet groups |
| Diagnostic delay (years) | 6.01 | 5.6 | ±2.0 | ✅ PASS | NORD 2020 |
| Mean misdiagnoses | 2.83 | 2.8 | ±1.0 | ✅ PASS | NORD 2020 |
| ACMG Pathogenic/LP rate | 77.9% | 80% | ±10% | ✅ PASS | ClinVar |
| ACMG VUS rate | 16.7% | 13% | ±6% | ✅ PASS | ClinVar |
| Inheritance pattern diversity | 6 | ≥5 | ±1 | ✅ PASS | Genetic Alliance |
| Trial template count | 15 | 15 | — | ✅ PASS | Schema invariant |
| Active arm responder rate | 71.1% | 65% | ±15% | ✅ PASS | Augustine et al. (2023) |
| Protocol deviation rate | 7.6% | 8% | ±3% | ✅ PASS | ICH GCP E6(R2) |

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

---

## Loading examples

### Pandas — explore the cohort

```python
import pandas as pd

cohort = pd.read_csv("rare_disease_cohort.csv", low_memory=False)

# Disease group distribution
print(cohort["disease_group"].value_counts())

# Diagnostic odyssey by disease group
print(cohort.groupby("disease_group")[
    ["diagnostic_delay_years", "n_misdiagnoses"]
].mean().round(2).sort_values("diagnostic_delay_years"))

# ACMG classification mix
print(cohort["acmg_classification"].value_counts(normalize=True).round(3))
```

### Trial template inspection

```python
import json
with open("trial_templates.json") as f:
    templates = json.load(f)

for trial_id, t in list(templates.items())[:3]:
    print(f"\n{trial_id} — {t['disease']} ({t['phase']})")
    print(f"  Intervention: {t['intervention']}")
    print(f"  N target: {t['n_target']}")
    print(f"  Primary endpoint: {t['primary_endpoint']}")
    print(f"  Required gene: {t['inclusion']['required_gene']}")
    print(f"  Age window: {t['inclusion']['min_age']}-{t['inclusion']['max_age']}")
```

### Per-trial screening funnel

```python
import pandas as pd

funnel = pd.read_csv("trial_screening_funnel.csv")
print(funnel.to_string(index=False))

# Trials meeting ≥80% target
print(f"\nTrials meeting target: {funnel['n_target_met'].sum()}/{len(funnel)}")
```

### Eligibility analysis

```python
import pandas as pd

elig = pd.read_csv("trial_eligibility.csv")
cohort = pd.read_csv("rare_disease_cohort.csv", low_memory=False)

# Eligibility rate by trial
print(elig.groupby("trial_id")["eligible_flag"].mean().sort_values(ascending=False))

# Distribution of eligibility scores
print(elig["eligibility_score"].describe())

# Top eligibility blockers per trial
enrich = pd.read_csv("trial_enrichment.csv")
print("\nTop blocker per trial:")
print(enrich[["trial_id", "top_blocker", "eligibility_rate"]])
```

### Endpoint tracking

```python
import pandas as pd

endpoints = pd.read_csv("trial_endpoints.csv")

# Responder rate by trial (active arms only)
active = endpoints[endpoints["active_drug_flag"] == 1]
print(active.groupby("trial_id")["patient_responder_flag"].mean().round(3))

# Protocol deviation rate by trial
print(endpoints.groupby("trial_id")["protocol_deviation_flag"].mean().round(3))
```

### Hugging Face Datasets

```python
from datasets import load_dataset

ds = load_dataset("xpertsystems/hlt011-sample", data_files={
    "cohort":          "rare_disease_cohort.csv",
    "eligibility":     "trial_eligibility.csv",
    "enrollment":      "trial_enrollment.csv",
    "endpoints":       "trial_endpoints.csv",
    "funnel":          "trial_screening_funnel.csv",
    "enrichment":      "trial_enrichment.csv",
})
print(ds)
```

---

## Suggested use cases

- **Trial recruitment optimization** — predict eligibility score from patient features; identify under-served eligible populations
- **Patient-to-trial matching ML** — train ranking models on `eligibility_score` × patient feature crosses
- **Screening funnel analytics** — analyze drop-off at each I/E criterion; build dropout-cause classifiers
- **Diagnostic odyssey prediction** — predict `diagnostic_delay_years` from symptom-onset features; identify patients at risk of delayed diagnosis
- **ACMG variant reclassification modeling** — train VUS-to-Pathogenic reclassification predictors
- **HPO-based phenotype clustering** — cluster patients by HPO term overlap for diagnostic ML
- **Treatment response prediction** — predict `response_status` from baseline biomarkers + PGx
- **Adverse event modeling** — predict CTCAE grade from treatment + comorbidities
- **QoL trajectory modeling** — model `qol_eq5d_baseline → qol_eq5d_12mo` change
- **Caregiver burden modeling** — predict ZBI score from disease severity + access factors
- **Rare disease registry simulation** — schema-compliant data for registry ETL testing
- **Patient advocacy targeting** — identify low-`patient_advocacy_flag` cohorts for outreach
- **Healthcare AI pretraining** — pretrain rare disease models before fine-tuning on real registry data (NORD, Genetic Alliance partner registries)

---

## Sample vs. full product

| Aspect | This sample | Full HLT-011 product |
|---|---|---|
| Patients | 1,500 | 50,000+ (default 5,000) up to 500K |
| Diseases | 50 catalog (48 represented at sample N) | 50 catalog (full coverage) |
| Trials | 15 templates | 15 templates (configurable) |
| Schema | identical (107 cohort cols + 6 trial CSVs) | identical |
| Calibration | identical | identical |
| License | CC-BY-NC-4.0 | Commercial license |

The full product unlocks:
- **Up to 500K patients** for production-grade rare disease ML training
- **Full disease catalog representation** including ultra-rare diseases that are statistically unlikely to sample at preview scale
- **Multi-year longitudinal extensions** (natural history, registry follow-up)
- **Custom trial template additions** beyond the 15 included designs
- Commercial use rights

**Contact us for the full product.**

---

## Limitations & honest disclosures

- **Sample is preview-only.** 1,500 patients × 48 diseases × 15 trials is enough to demonstrate schema, calibration, and the full trial-eligibility engine output, but is **not statistically sufficient** for per-disease ML modeling (most diseases have ~30 patients in the sample). Use the full product (50K+) for serious work.
- **2 of 50 diseases are not represented in this sample.** Ultra-rare diseases (prevalence <0.1/100k) may not be sampled at n=1500. The full product hits all 50 diseases at scale.
- **`ultra_rare_flag` rate runs low (~0.001 vs config target 0.15).** The catalog includes ultra-rare diseases, but the patient-level sampling weights by prevalence — so common rare diseases dominate at this N. To get ultra-rare-enriched cohorts, use the full product with `ULTRA_RARE_PCT=0.15` parameter.
- **Trial enrollment ratios reflect rare disease realities.** Most trials enroll 25-50% of their N target during the 24-month sample window because rare diseases have small eligible populations. The full product (50K patients) achieves target enrollment for nearly all trials.
- **Eligibility scoring is composite-based, not adjudicated.** Real trial screening involves clinician judgment beyond the 10 weighted criteria. Use the scores for ML pipeline development; do not interpret as actual recruitment decisions.
- **Variant HGVS strings are synthetic.** `variant_hgvs_cdna` and `variant_hgvs_protein` follow HGVS nomenclature format but are computationally generated, not pulled from ClinVar.
- **HPO terms are sampled from disease-group-specific pools, not patient-specific phenotype mapping.** Real HPO annotation comes from manual deep phenotyping. The sample includes realistic HPO term *counts* and *distributions* but the specific term-to-patient mapping is statistical.
- **Trial responder rates run slightly high (71% vs 60-65% typical).** Rare disease mechanism-based therapies do tend toward high response rates compared to common-disease drugs (~30-50% common-disease responder), so 71% is plausible but on the high end.
- **No real ClinVar IDs, ORPHA codes are accurate to catalog, OMIM IDs accurate.** ORPHA codes and OMIM IDs reference real entries in those databases. The variant-level `clinvar_id` field uses synthetic placeholder IDs.
- **Synthetic, not derived from real rare disease patient cohorts.** Distributions match published NORD/ORPHA/ClinVar references but do NOT reflect any specific real cohort.

---

## Ethical use guidance

This dataset is designed for:
- Rare disease ML methodology development
- Clinical trial recruitment optimization research
- Diagnostic odyssey reduction methodology
- ACMG variant classification ML
- HPO-based phenotype clustering research
- Healthcare AI pretraining for rare disease prediction tasks
- Educational use in medical genetics, clinical trial design, and rare disease epidemiology

This dataset is **not appropriate for**:
- Making clinical decisions about real patients with rare diseases
- Variant pathogenicity calls without ACMG framework validation
- Trial recruitment decisions for real patients without IRB/clinician oversight
- FDA submissions for orphan drug development
- Discriminatory analyses targeting protected demographic groups or rare disease patient populations

**Note on rare disease patient communities**: Rare disease patients and families are uniquely vulnerable. Synthetic data MUST NOT be used in ways that could be interpreted as representing or speaking for real patient experiences. Use this dataset for technical/methodological work only.

---

## 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)
- [HLT-008](https://huggingface.co/datasets/xpertsystems/hlt008-sample) — Synthetic Healthcare Claims Dataset (500 members + 30K claims + fraud)
- [HLT-009](https://huggingface.co/datasets/xpertsystems/hlt009-sample) — Synthetic Continuous Vital Sign Monitoring Dataset (25 ICU episodes)
- [HLT-010](https://huggingface.co/datasets/xpertsystems/hlt010-sample) — Synthetic Hospital Resource Usage Dataset (OR + Staffing + Equipment)
- **HLT-011** — Synthetic Rare Disease + Trial Eligibility Engine (you are here)

Use **HLT-001 through HLT-011 together** for the full healthcare data stack — and HLT-011 specifically extends the catalog into **rare disease & orphan drug development**, complementing HLT-003 (clinical trial design) and HLT-007 (pharmacology) with rare-disease-specific patient phenotyping and trial recruitment workflows.

---

## Citation

If you use this dataset, please cite:

```bibtex
@dataset{xpertsystems_hlt011_sample_2026,
  author       = {XpertSystems.ai},
  title        = {HLT-011 Synthetic Rare Disease Dataset + Clinical Trial Eligibility Engine (Sample Preview)},
  year         = 2026,
  publisher    = {Hugging Face},
  url          = {https://huggingface.co/datasets/xpertsystems/hlt011-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.