xpertsystems/hc-gen-001-sample

HC-GEN-001 — Pulmonary Fibrosis: Rare Genetic Disease, Pulmonary Function, Infection & Gene Therapy Response (Sample)

Synthetic patient-level dataset for idiopathic and genetic pulmonary fibrosis (PF/ILD), spanning telomere/surfactant-protein genetics, longitudinal pulmonary function, respiratory infection burden, AAV gene-therapy pharmacodynamics, antifibrotic therapy, acute exacerbations, biomarker trajectories, and clinical outcomes.

This is a 500-patient sample of the full HC-GEN-001 product. It is synthetic — generated by a calibrated simulation engine anchored to published cohort and trial statistics. It contains no real patient data.

Not for clinical use. This dataset is for ML development, benchmarking, schema prototyping, and education only. It must not be used to inform real patient care, diagnosis, or treatment decisions.

Calibration anchors

The generator is calibrated so cohort-level statistics fall within ranges drawn from the literature. Sample-level observed values (seed 42, n=500):

Metric	Observed	Target	Anchor
IPF subtype prevalence	0.63	0.60–0.69	ATS/ERS/JRS/ALAT 2022 Guideline; IPFnet
Familial PF prevalence	0.18	0.11–0.20	Garcia-Sancho 2011; PROFILE
Male sex fraction	0.56	0.52–0.63	IPFnet / INPULSIS / ASCEND
MUC5B carrier rate	0.41	0.36–0.50	Seibold et al. NEJM 2011
Definite-UIP HRCT rate	0.53	0.50–0.60	ATS/ERS/JRS/ALAT 2022
FVC % predicted (mean)	67.4	62–72	CAPACITY / ASCEND
DLCO % predicted (mean)	52.7	47–57	INPULSIS
Acute exacerbation rate	0.25	0.20–0.32	Collard 2016; PROFILE
Mortality (follow-up)	0.43	0.33–0.55	ISHLT Registry / IPFnet
Age onset gap (IPF − FPF)	13.3 yr	≥8 (floor)	Telomeropathy earlier onset (Armanios)
Telomere separation (FPF/DC)	1.69 kb	≥0.7 (floor)	Alder / Armanios telomere biology
FVC−DLCO restrictive gap	14.7 pp	≥10 (floor)	ATS/ERS PFT physiology
FVC decline separation (FPF)	34.1 mL	≥15 (floor)	PROFILE accelerated decline
GT treatment−placebo FVC₆ₘ	185.6 mL	≥60 (floor)	AAV gene-therapy response separation

Validation: Grade A+ (10.00/10) across all six canonical seeds (42, 7, 123, 2024, 99, 1), deterministic.

Schema highlights by module

162 columns, single table. Module groups:

• Demographics & diagnosis — pf_subtype (IPF / FPF / HPS_PF / DC_PF / SP_Mutation_PF), age_at_diagnosis, upa_pattern, hrct_fibrosis_extent_pct, honeycombing_flag, gapindex_score, diagnostic_delay_months, facility_type.
• Genetics — telomere_gene_mutation (TERT/TERC/RTEL1/PARN/NAF1/TINF2), tert_variant_class, telomere_length_kb, surfactant_protein_mutation (SFTPC/SFTPA1/SFTPA2/ABCA3), sftpc_variant, muc5b_promoter_rs35705950, variant_classification, zygosity, penetrance_modifier.
• Pulmonary function — fvc_pct_predicted_baseline, dlco_pct_predicted_baseline, tlc_pct_predicted_baseline, fev1_fvc_ratio_baseline, sixmwt_meters_baseline, spo2_nadir_6mwt_pct, pulmonary_hypertension_flag, rvsp_mmhg, annual decline rates, and a 41-point quarterly FVC trajectory (fvc_quarterly_trajectory_pct, JSON string).
• Infection — on_immunosuppression_flag, bacterial/viral episode counts, pjp_episode_flag, aspergillus_infection_flag + species/antifungal, influenza/SARS-CoV-2 flags, BAL cellular profile, vaccination flags.
• Gene therapy (AAV) — gene_therapy_enrollment_flag, gene_therapy_product, gene_therapy_arm, vector_serotype, delivery_route, vector_dose_vg_kg, baseline_nab_titer_aav, transduction_efficiency_pct, FVC/DLCO response at 3m/6m, gt_responder_flag, immune/CRS/SAE flags, transgene_persistence_months.
• Antifibrotic therapy — antifibrotic_drug (pirfenidone/nintedanib), dosing, dose_reduction_flag, treatment_discontinuation_flag + reason, GI/photosensitivity/hepatotoxicity AEs, adherence_antifibrotic_pct, transplant referral/completion.
• Acute exacerbation — ipf_ae_flag, trigger, severity, ICU/ventilation, hospitalization days, ae_mortality_flag, HRCT pattern, time from diagnosis.
• Biomarkers — SP-D, KL-6, MMP-7, CCL18, CA-125, periostin, telomere dysfunction foci, RNA-seq (TGFβ1/WNT5A/SPP1), 4-marker proteomic composite.
• Outcomes & QoL — mortality_flag + cause + time, hospitalization, GAP-index 3yr change, SF-36 PCS, UCSD SOBQ, EQ-5D, PHQ-9, GAD-7, palliative care flags.
• Coding standards — ICD-11, SNOMED-CT, LOINC (FVC/DLCO), OMOP, HPO.

Files

• hc_gen_001_sample.csv — 500 synthetic patients, 162 columns.
• README.md — this file.

Loading

import pandas as pd
df = pd.read_csv("hc_gen_001_sample.csv")

# Decode the longitudinal FVC trajectory
import json
df["fvc_traj"] = df["fvc_quarterly_trajectory_pct"].apply(json.loads)
print(df.shape)  # (500, 162)

from datasets import load_dataset
ds = load_dataset("xpertsystems/hc-gen-001-sample")

Use cases

• Rare-disease / ILD subtyping and risk-stratification model development.
• Gene-therapy trial-design simulation (arm separation, responder modeling, immunogenicity).
• Longitudinal FVC/DLCO decline forecasting.
• Survival and competing-risk modeling (mortality vs transplant vs AE).
• Synthetic-data methodology benchmarking and schema/ETL prototyping (OMOP mapping).

Limitations (honestly disclosed)

• Synthetic, cross-sectional-with-embedded-longitudinal design. Each row is one patient; longitudinal structure is encoded in summary rates plus the fvc_quarterly_trajectory_pct JSON array, not as separate visit rows.
• Marginal calibration, not full joint fidelity. Univariate prevalences and the engineered structural separations (subtype age/telomere/decline gaps, GT arm separation) are anchored to literature; higher-order correlations beyond those engineered are not independently validated.
• Gene-therapy module is forward-looking/illustrative. AAV PF gene therapy is largely investigational; pharmacodynamic parameters represent plausible trial-design values, not observed approved-product data.
• Subtype-restricted GT enrollment. Enrollment is forced to 0 for HPS_PF and DC_PF subtypes, so the effective enrollment rate is slightly below the nominal gene_therapy_prevalence; the sample observes ~0.17 against a 0.20 nominal. This is intentional eligibility gating, not a calibration miss.
• No engine bugs identified. The HC-GEN-001 engine passes its own internal 14/14 validation across all six canonical seeds with stable 162-column schema; the Weibull sampler uses the correct inverse-CDF form. No defects were observed or suppressed.
• Small-sample variance. At n=500 some categorical rates (FPF, MUC5B, Aspergillus, AE) carry sampling variance; scorecard ranges accommodate this without masking real misses, and structural floors are weighted to dominate.

Commercial / full version

	Sample (this)	Full (commercial)
Patients	500	20,000+ (configurable)
Columns	162	162
Longitudinal	Embedded trajectory	Full visit-level expansion available
Seeds / reproducibility	6 canonical	Unlimited
Formats	CSV	CSV / JSON / Parquet
License	CC-BY-NC-4.0	Commercial
Support	—	SLA, custom calibration, vertical extensions

Contact pradeep@xpertsystems.ai · https://xpertsystems.ai

Citation

@dataset{xpertsystems_hcgen001_2026,
  title  = {HC-GEN-001: Synthetic Pulmonary Fibrosis Dataset --- Rare Genetic
            Disease, Pulmonary Function, Infection & Gene Therapy Response (Sample)},
  author = {XpertSystems.ai},
  year   = {2026},
  publisher = {Hugging Face},
  note   = {Synthetic data. Not for clinical use. Calibration anchors:
            ATS/ERS/JRS/ALAT 2022 IPF Guideline (Raghu et al.);
            Seibold et al. NEJM 2011 (MUC5B rs35705950);
            CAPACITY (Noble et al. 2011); ASCEND (King et al. 2014);
            INPULSIS (Richeldi et al. 2014); PROFILE Cohort;
            IPFnet; ISHLT Thoracic Transplant Registry;
            Garcia-Sancho et al. 2011 (familial IPF);
            Armanios / Alder telomere biology; Collard et al. 2016 (AE-IPF).},
  url    = {https://huggingface.co/datasets/xpertsystems/hc-gen-001-sample}
}