A comprehensive synthetic cohort of Polycystic Ovary Syndrome patients organized around the Rotterdam phenotypes (A/B/C/D) and spanning HPO-axis hormonal profiles (LH/FSH, androgens, AMH), insulin resistance & metabolic syndrome, ovarian morphology (AFC, volume, endometrium), fertility outcomes (letrozole/clomiphene/IUI/IVF with OHSS risk), pharmacological treatment trajectories, and long-term cardiometabolic risk. This repository contains a 500-row, single-seed sample. The full commercial product scales to 20,000+ patients with 15-year follow-up trajectories and CSV / Parquet / JSON / FHIR R4 delivery.

SKU: HC-END-008
Sample size: 500 patients × 190 columns
License (sample): CC-BY-NC-4.0 — commercial license available for the full product
Contact: pradeep@xpertsystems.ai · https://xpertsystems.ai

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 — including PCOS physiology checks (hyperandrogenic phenotypes carry higher testosterone). This engine also passes its own built-in 9-benchmark suite (Azziz phenotype distribution, PPCOS-II letrozole live-birth, Rotterdam AFC, AMH elevation, HOMA-IR, metabolic syndrome, SART IVF oocytes, depression, LH/FSH).

Calibration anchors

Metric	Sample value	Target range	Source
Phenotype A fraction	34.2%	31–45%	Azziz 2016 (Rotterdam A ~33-43%)
LH/FSH ratio mean	2.66	1.5–3.5	PCOS LH hypersecretion
Elevated AMH fraction	77.0%	40–82%	PCOS AMH elevation
Mean HOMA-IR	3.97	2.4–4.5	PCOS insulin resistance
Insulin resistance (HOMA≥2.5)	76.0%	65–85%	PCOS IR prevalence
Metabolic syndrome	42.4%	30–50%	PCOS MetS prevalence
Letrozole live-birth rate	27.1%	24–31%	PPCOS-II (Legro 2014 NEJM ~27.5%)
Depression	36.8%	28–42%	PCOS depression prevalence
Testosterone separation (HA−nonHA)	30.6 ng/dL	≥15	Hyperandrogenism physiology
AFC in plausible [4,70]	100%	≥1.0	Ovarian morphology bounds
TSH in [0.01,10]	100%	≥1.0	Assay-range integrity
Column count	190	≥184	Schema completeness (9 modules)

Schema highlights by module (190 columns)

Demographics. Age, race (incl. South/East Asian PCOS-relevant strata), height/weight/BMI, insurance, region.

Phenotype & diagnosis. Rotterdam A/B/C/D, hyperandrogenism / oligo-anovulation / PCO-morphology flags, menstrual pattern & cycle length, Ferriman-Gallwey, hirsutism/acne/alopecia/acanthosis, diagnosis age & delay.

Hormonal profile. LH/FSH & ratio, GnRH pulse frequency, estradiol/estrone/progesterone, total/free testosterone, SHBG, DHEAS, androstenedione, 17-OHP, DHT, AMH, inhibin B, prolactin, TSH/FT4, cortisol/ACTH; annual LH/testosterone/AMH JSON trajectories; elevation flags.

Insulin resistance & metabolic. Fasting glucose/insulin, HOMA-IR/B, QUICKI, OGTT, HbA1c, glycemic category, adipokines (adiponectin/leptin/ghrelin), C-peptide, VAI, waist/WHR, full lipid panel, inflammatory markers, BP, MetS criteria, NAFLD, sleep apnea, HOMA-IR trajectory.

Ovarian morphology. AFC (R/L/total), ovarian volume, dominant follicle, stromal ratio & echogenicity, endometrial thickness/pattern/hyperplasia/cancer, fibroids, hemorrhagic cyst, Rotterdam AFC & volume criteria.

Fertility. Intent, infertility duration, ovulation induction (letrozole/clomiphene/FSH/GnRH), PPCOS-II-calibrated ovulation & live-birth rates, IUI, IVF (oocytes/fertilization/blastocyst/CPR/ LBR), FET, OHSS risk & events, pregnancy losses, GDM/preeclampsia/preterm/NICU, time-to-conception.

Pharmacotherapy. OCP (progestin type, testosterone reduction, SHBG increase), metformin, spironolactone, GLP-1 RA, inositol, laser/electrolysis, adherence, response, switching.

Cardiometabolic & long-term risk. 10yr CVD risk, events, hypertension onset, carotid IMT, coronary calcium, endothelial dysfunction, endometrial/ovarian/breast cancer, depression/anxiety/ eating disorder, PCOSQ quality of life, sexual dysfunction, body-image distress.

Clinical labs & utilization. Vitamin D, ferritin, hemoglobin, LFTs, eGFR, uric acid, heart rate; visit/referral counts, ultrasound/lab counts, fertility & total costs.

Coding. FHIR R4 Patient bundle (full product).

Files

hc_end_008_sample.csv — 500-patient sample (190 columns)
generate_sample_dataset_hc_end_008.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, json
df = pd.read_csv("hc_end_008_sample.csv")
print(df[["patient_id","pcos_phenotype_rotterdam","homa_ir",
          "anti_mullerian_hormone_ng_ml","ovulation_induction_type"]].head())
# Annual trajectories are JSON-encoded:
amh_traj = json.loads(df.loc[0, "amh_trajectory_annual_json"])

from datasets import load_dataset
ds = load_dataset("csv", data_files="hc_end_008_sample.csv")

Use cases

PCOS phenotype classification (Rotterdam A/B/C/D from hormone + morphology features)
Fertility-outcome modeling (ovulation induction & IVF response prediction)
Insulin-resistance and metabolic-syndrome risk stratification
OHSS-risk prediction tooling
Treatment-response and comparative-effectiveness modeling
ML training where real PCOS reproductive-endocrine EHR data is PHI-restricted

Honest limitations & disclosed generator behavior

This is a well-calibrated engine (passes its own 9-benchmark suite, correct PCOS physiology). The following are standard caveats and minor specifics:

Independent symptom/flag draws. Many clinical flags (acne, alopecia, acanthosis, comorbidity flags) are drawn independently conditioned on phenotype, so within-patient symptom clustering is weaker than in real cohorts. Phenotype-level prevalences are correct.
Annual trajectories are simplified. The LH/testosterone/AMH/HOMA-IR JSON arrays are 15-point series generated as baseline + small annual noise (AMH with a mild downward drift); they do not model treatment-driven inflections or pregnancy-related excursions.
Fertility outcomes are marginal rates. Letrozole/clomiphene/IVF success fields are trial-calibrated population rates rather than individual-conditioned outcomes tied to that patient's exact AMH/AFC/age jointly.
Some exclusion conditions are overlays. Hyperprolactinemia and hypothyroidism are applied as independent ~10-12% overlays (as PCOS-mimic exclusions) rather than fully separated diagnoses.

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-008 product
Patients	500	20,000+ (configurable)
Follow-up	Baseline + 15pt JSON trajectories	Full 15-year longitudinal panel
Seeds / cohorts	1	Multi-seed, reproducible
Formats	CSV	CSV + Parquet + JSON + FHIR R4 Bundle
Symptom clustering	Independent draws	Correlated within-patient symptom model
Fertility outcomes	Marginal trial rates	Individual-conditioned joint model
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_008_2026,
  title        = {HC-END-008: PCOS 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: Azziz et al. 2016
                  (PCOS phenotype epidemiology); PPCOS-II / Legro et al. 2014 NEJM
                  (letrozole vs clomiphene for infertility in PCOS); ESHRE/ASRM Rotterdam
                  consensus criteria; SART national IVF outcomes registry; international
                  evidence-based PCOS guideline (Teede et al. 2018/2023).}
}

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

Downloads last month: 20

Total file size:

605 kB