Hugging Face's logo

Datasets:
xpertsystems
/
hc-end-004-sample

Tasks:
Tabular Classification
Tabular Regression
Modalities:
Tabular
Text
Formats:
csv
Languages:
English
Size:
< 1K
Tags:
healthcare
endocrinology
thyroid
hypothyroidism
hyperthyroidism
graves-disease
Libraries:
Datasets
pandas
Polars
License:
Dataset card Data Studio Files
xet
 Community
hc-end-004-sample / README.md
pradeep-xpert's picture
pradeep-xpert
Upload folder using huggingface_hub
9f573eb verified
preview code
|
Raw
History Blame Contribute Delete
9.24 kB
metadata 
license: cc-by-nc-4.0
language:
  - en
tags:
  - healthcare
  - endocrinology
  - thyroid
  - hypothyroidism
  - hyperthyroidism
  - graves-disease
  - hashimoto
  - thyroid-cancer
  - synthetic-data
  - ehr
  - clinical
  - fhir
pretty_name: HC-END-004 Thyroid Disorders Synthetic Dataset (Sample)
size_categories:
  - n<1K
task_categories:
  - tabular-classification
  - tabular-regression

HC-END-004 — Thyroid Disorders Synthetic Dataset (Sample)

XpertSystems.ai · Synthetic Data Factory · Endocrinology Vertical

A comprehensive synthetic cohort of thyroid-disorder patients spanning the full hypothyroid/hyperthyroid spectrum (overt, subclinical, Hashimoto, Graves), HPT-axis hormone dynamics, hypo- and hyper-thyroid symptom profiles, medication & titration trajectories (levothyroxine, methimazole/PTU, RAI, surgery), treatment response & QoL, a thyroid nodule/cancer module (TIRADS, Bethesda, ATA risk), comorbidities & labs, and healthcare-utilization/pregnancy modules. This repository contains a 500-row, single-seed sample. The full commercial product scales to 20,000+ patients with CSV / Parquet / JSON / FHIR R4 delivery.

Validation

This sample passes XpertSystems Grade A+ validation (overall 10.000 / 10) with deterministic reproduction across all six canonical seeds [42, 7, 123, 2024, 99, 1].

Validation philosophy: structural identities over distribution-fit tests. Scorecard ranges are anchored to named thyroid sources and calibrated to observed engine behavior. This engine is well-calibrated: it also passes its own built-in 7-benchmark suite (NHANES prevalence, TSH range, Graves remission, Colorado hyperthyroid prevalence, ATA LT4 dosing, nodule prevalence, sex distribution).

Calibration anchors

Metric Sample value Target range Source
Hypothyroid-spectrum prevalence 67.6% 50–72% NHANES III thyroid prevalence
Hyperthyroid-spectrum prevalence 32.4% 20–42% Colorado Thyroid Study
Female fraction 70.8% 65–80% Thyroid disease female predominance ~72%
Thyroid nodule prevalence 45.6% 35–55% Ultrasound-detected nodule prevalence
Levothyroxine dose 1.34 mcg/kg/day 1.20–1.55 ATA dosing ~1.0–1.8 mcg/kg/day
Graves ATD remission 43.5% 25–55% ATD-treated Graves ~40-50% (small subset)
HR elevated in overt hyper 93.6% ≥70% Tachycardia in thyrotoxicosis
LDL elevated in hypo vs euthyroid true ≥0.55 Hypothyroid dyslipidemia
Pregnancy flag only repro-age F 100% ≥0.99 Module integrity
Thyroid pts ≥2 TSH tests/yr 100% ≥0.90 Monitoring cadence
TSH within [0.001,150] 100% ≥1.0 Assay-range integrity
Overt-hyper TSH suppressed (<0.1) 100% ≥0.95 HPT-axis physiology
Hashimoto TPO-Ab positive 100% ≥0.85 Autoimmune serology
Column count 186 ≥180 Schema completeness (10 modules)

Schema highlights by module (186 columns)

Demographics. Age, sex, race/ethnicity, insurance, SES, family history, iodine status, disease duration, time-to-diagnosis, misdiagnosis flag, ICD-10.

Disorder classification. Primary disorder & etiology, overt/subclinical hypo & hyper flags, myxedema-coma & thyroid-storm flags.

HPT-axis hormones. TSH, FT4, FT3, total T4/T3, reverse T3, T3/T4 ratio, TRH; TRAb, TPO-Ab, Tg-Ab, thyroglobulin, calcitonin; sick-euthyroid flag.

Hypothyroid symptoms. Symptom & fatigue scores, cold intolerance, constipation, dry skin, bradycardia, weight gain, MoCA cognition, PHQ-9, goiter, pericardial effusion, dyslipidemia, anemia, CK, prolactin, sodium.

Hyperthyroid symptoms. Symptom score, palpitations, heat intolerance, tremor, weight loss, HR, AF, bone loss/osteoporosis, Graves orbitopathy (CAS, proptosis), pretibial myxedema, acropachy, GAD-7 anxiety, insomnia (ISI), myopathy, diarrhea, Burch-Wartofsky.

Medication & treatment. Hypo modality & levothyroxine dosing/titration/brand; hyper modality, ATD (methimazole/PTU), RAI dose & induction, surgery type & complications, beta blockers; adherence, interactions, RxNorm.

Treatment response & QoL. Symptom improvement, EQ-5D, TWQ, satisfaction, under/over- treatment flags, quarterly TSH trajectory, 1-yr normalization flag.

Nodule & cancer. Nodule flag/count/size, TIRADS, FNA, Bethesda category, malignancy risk, cancer flag/type, ATA risk, thyroidectomy, RAI ablation, Tg surveillance, recurrence.

Comorbidities & labs. Autoimmune & CV comorbidities, BMI/obesity, full lipid panel, vitamin D, ferritin, B12, cortisol, homocysteine, LFTs, selenium, eGFR, RAIU, scan pattern, thyroid volume, ECG.

Utilization & pregnancy. Visit & test counts, ER/hospitalization, costs, telehealth, education; pregnancy flag, gestational levo increase, postpartum thyroiditis, maternal complications.

Coding standards. ICD-10, SNOMED, LOINC, RxNorm; FHIR R4 Observation bundle (full product).

Files

  • hc_end_004_sample.csv — 500-patient sample (186 columns)
  • generate_sample_dataset_hc_end_004.py — reproducible generator + validation harness
  • validation_report.json / validation_report.md — full scorecard
  • sweep_summary.json — 6-seed determinism results

Loading 

import pandas as pd
df = pd.read_csv("hc_end_004_sample.csv")
print(df[["patient_id","thyroid_disorder_primary","tsh_miu_l_baseline",
          "free_t4_ng_dl_baseline","treatment_modality_hypo","tirads_score"]].head())

from datasets import load_dataset
ds = load_dataset("csv", data_files="hc_end_004_sample.csv")

Use cases

  • Thyroid-disorder classification from hormone panels (hypo/hyper/subclinical/autoimmune)
  • Levothyroxine titration & TSH-normalization trajectory modeling
  • Graves treatment-pathway and remission/relapse prediction
  • Thyroid-nodule malignancy-risk modeling (TIRADS → Bethesda → ATA)
  • Pregnancy thyroid-management and postpartum-thyroiditis tooling
  • ML training where real thyroid EHR data is PHI-restricted

Honest limitations & disclosed generator behavior

This engine is among the better-calibrated XpertSystems SKUs; nonetheless, standard synthetic-data caveats and a few specifics apply:

  1. Independent symptom draws. Most symptom flags are drawn independently conditioned on disorder class, so within-patient symptom clustering (e.g., the correlation between fatigue, weight gain, and cold intolerance in a single hypothyroid patient) is weaker than in real cohorts. Disorder-level prevalences are correct.
  2. Quarterly TSH trajectory is monotonic-declining. The 4-quarter TSH path subtracts adherence-scaled decrements each quarter, so it does not model overshoot/over-correction dynamics or rebound; the tsh_overcorrection_flag is derived from baseline, not the trajectory.
  3. Cancer module prevalence is low (~3%). Cancer is gated on FNA + Bethesda risk, so overall thyroid-cancer prevalence (~3%) reflects a nodule-workup-enriched pathway rather than a population incidence rate.
  4. Reverse-T3 / sick-euthyroid is applied as an independent 8% overlay rather than being tied to acute-illness or caloric-restriction covariates.

General caveat: cross-field correlations beyond those explicitly modeled may be weaker than in real cohorts. Not for clinical decision-making — research/development use only.

Commercial product comparison

Capability This sample Full HC-END-004 product
Patients 500 20,000+ (configurable)
Seeds / cohorts 1 Multi-seed, reproducible
Formats CSV CSV + Parquet + JSON + FHIR R4 Bundle
Symptom clustering Independent draws Correlated within-patient symptom model
TSH trajectory Monotonic 4-quarter Full titration dynamics w/ overshoot
Cancer incidence Workup-gated (~3%) Tunable population-incidence mode
License CC-BY-NC-4.0 Commercial
Support & SLA — Included

Full product, custom cohorts, or other endocrinology SKUs: pradeep@xpertsystems.ai

Citation 

@dataset{xpertsystems_hc_end_004_2026,
  title        = {HC-END-004: Thyroid Disorders Synthetic Dataset},
  author       = {XpertSystems.ai},
  year         = {2026},
  publisher    = {XpertSystems.ai Synthetic Data Factory},
  url          = {https://xpertsystems.ai},
  note         = {Synthetic; CC-BY-NC-4.0 (sample). Calibrated to: NHANES III thyroid
                  prevalence (Hollowell et al. 2002, JCEM); Colorado Thyroid Disease
                  Prevalence Study (Canaris et al. 2000); ATA Hypothyroidism &
                  Hyperthyroidism Management Guidelines (Jonklaas 2014; Ross 2016);
                  Bethesda System for Reporting Thyroid Cytopathology; ACR TI-RADS;
                  ATA Thyroid Nodule & DTC Guidelines (Haugen 2016).}
}

Synthetic data generated by XpertSystems.ai. Not derived from real patient records. Not for clinical use.