Upload folder using huggingface_hub

README.md

@@ -0,0 +1,210 @@
+---
+license: cc-by-nc-4.0
+language:
+- en
+tags:
+- healthcare
+- respiratory
+- pulmonology
+- pneumonia
+- cap
+- hap
+- vap
+- sepsis
+- curb-65
+- psi-score
+- antibiotics
+- synthetic-data
+- ehr
+- clinical
+pretty_name: "HC-RES-003 Pneumonia Synthetic Dataset (Sample)"
+size_categories:
+- n<1K
+task_categories:
+- tabular-classification
+- tabular-regression
+- survival-analysis
+---
+
+# HC-RES-003 — Pneumonia Synthetic Dataset (Sample)
+
+**XpertSystems.ai · Synthetic Data Factory · Respiratory Vertical**
+
+The third SKU in the XpertSystems **Respiratory** vertical: a clinically detailed synthetic
+pneumonia cohort spanning **Community-Acquired (CAP)**, **Hospital-Acquired (HAP)**,
+**Ventilator-Associated (VAP)**, and **Healthcare-Associated (HCAP)** pneumonia. It covers organism
+taxonomy & resistance (MRSA/MDR), full severity scoring (CURB-65, PSI/Fine, SMART-COP, SOFA,
+APACHE-II, NEWS2, ARDS Berlin), vitals, labs & ABG, respiratory support (HFNC/NIV/IMV/ECMO,
+lung-protective ventilation), guideline-based antibiotic regimens, comorbidities, and
+hospitalization outcomes (ICU, mortality, complications, readmission, cost).
+
+This repository contains a **500-row, single-seed sample**. The full commercial product scales to
+20,000+ patients with CSV / Parquet / JSON delivery.
+
+- **SKU:** HC-RES-003
+- **Sample size:** 500 patients × 157 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 the
+VAP > CAP mortality gradient and the procalcitonin bacterial-vs-viral separation as structural
+floors. This engine also passes its own built-in ~15-check benchmark suite (CAP/HAP/VAP
+distribution, PSI Class-V mortality, CURB-65=0 mortality, ICU/intubation rates, ARDS PF ratio).
+
+### Calibration anchors
+
+| Metric | Sample value | Target range | Source |
+|---|---|---|---|
+| CAP prevalence | 67.4% | 60–71% | ATS/IDSA 2019 (~60-70%) |
+| HAP prevalence | 22.0% | 18–28% | Hospital-acquired epidemiology |
+| VAP prevalence | 7.2% | 5–11% | Ventilator-associated epidemiology |
+| ICU admission rate | 21.8% | 15–30% | ATS/IDSA (~18-26%) |
+| Intubation rate | 22.2% | 14–27% | VAP-adjusted |
+| 30-day mortality | ~8.8% | 5–12% | Overall pneumonia mortality |
+| 30-day readmission | ~16.5% | 13–24% | CMS HRRP (~18%) |
+| Median hospital LOS | 4.1 days | 3.5–8 | CAP/HAP LOS literature |
+| **VAP − CAP mortality gradient** | **+0.11** | **≥0.03** | Severity ordering (physiology) |
+| **VAP > CAP mortality holds** | **yes** | **boolean** | Severity ordering |
+| **PCT bacterial / viral ratio** | **23.5×** | **≥3×** | Procalcitonin biomarker separation |
+| **Column count** | **157** | **≥150** | Schema completeness (10 modules) |
+
+---
+
+## Schema highlights by module (157 columns)
+
+**Demographics.** Bimodal age (pediatric + adult/elderly), sex, race, insurance, region, nursing-home
+residence, prior 90-day hospitalization.
+
+**Classification.** Pneumonia type (CAP/HAP/VAP/HCAP), etiology category, causative organism (type-
+specific taxonomies), MRSA & MDR flags, ICD-10 & SNOMED, chest X-ray finding, procalcitonin, CRP,
+WBC, blood culture, urine antigen, time-to-antibiotics.
+
+**Vitals.** Temperature, HR, RR, SBP/DBP/MAP, SpO2, GCS, BMI, plus fever/hypotension/hypoxemia/
+tachypnea flags.
+
+**Severity scores.** CURB-65 (+ component flags), PSI/Fine score & risk class, SMART-COP, SOFA (6
+subscores + total), APACHE-II, NEWS2, PF ratio, ARDS Berlin category, sepsis & septic-shock flags,
+ATS major criteria, final severity.
+
+**Laboratory.** CBC, neutrophil/lymphocyte ratio, electrolytes, BUN/urea, creatinine, glucose,
+albumin, lactate, D-dimer, ferritin, IL-6, troponin, plus AKI/thrombocytopenia/hyponatremia/
+hyperglycemia flags.
+
+**ABG.** FiO2, PaO2, PaCO2, pH, HCO3, lactate, SaO2.
+
+**Respiratory support.** O2 delivery device (Room Air → IMV-ECMO), intubation timing, ventilator
+days, lung-protective settings (tidal volume, PEEP, driving/plateau pressure), prone positioning,
+ECMO, NIV, HFNC, spontaneous breathing trial, extubation success, reintubation.
+
+**Antibiotics.** Guideline-based empiric regimen, primary & secondary agents (dose/freq/route/
+duration), PCT-guided de-escalation, oral switch, corticosteroids (indication, dexamethasone dose),
+antiviral/antifungal, vasopressors, C. diff, serial procalcitonin.
+
+**Comorbidities.** COPD, heart failure, diabetes, CKD, CAD, immunosuppression, HIV, malignancy,
+smoking, alcohol, frailty, vaccinations, aspiration risk.
+
+**Outcomes.** ICU admission & LOS, hospital LOS, inpatient/30d/90d mortality, time to death,
+complications (pleural effusion, empyema, AKI, new AF, bacteremia), discharge disposition, 30/90-day
+readmission, time to clinical stability, total cost.
+
+---
+
+## Files
+
+- `hc_res_003_sample.csv` — 500-patient sample (157 columns)
+- `generate_sample_dataset_hc_res_003.py` — reproducible generator + validation harness
+- `validation_report.json` / `validation_report.md` — full scorecard
+- `sweep_summary.json` — 6-seed determinism results
+
+## Loading
+
+```python
+import pandas as pd
+df = pd.read_csv("hc_res_003_sample.csv")
+print(df[["patient_id","pneumonia_type","severity_final","curb65_score",
+          "psi_risk_class","icu_admission_flag","mortality_30d_flag"]].head())
+```
+
+```python
+from datasets import load_dataset
+ds = load_dataset("csv", data_files="hc_res_003_sample.csv")
+```
+
+## Use cases
+
+- Pneumonia severity triage (CURB-65 / PSI / SMART-COP score replication & calibration)
+- Mortality and ICU-admission risk prediction
+- Antibiotic-stewardship and PCT-guided de-escalation modeling
+- Bacterial-vs-viral etiology classification from biomarkers
+- ARDS / respiratory-failure progression and ventilation-outcome modeling
+- Readmission-risk and cost modeling
+- ML training where real pneumonia EHR data is PHI-restricted
+
+---
+
+## Honest limitations & disclosed generator behavior
+
+This is a strongly calibrated single-table engine (passes its own ~15-check suite; correct severity
+and biomarker gradients). Standard caveats:
+
+1. **Cross-sectional snapshot.** Unlike the asthma (HC-RES-001) and COPD (HC-RES-002) SKUs, this is a
+   single admission-level table — there is no longitudinal trajectory or per-event companion table.
+   Each row is one pneumonia episode.
+2. **Severity proxy drives correlated modules.** Vitals, labs, and ABG are generated from a single
+   0–1 severity proxy (seeded by pneumonia type), so they co-move cleanly with severity but encode
+   somewhat tighter within-patient coupling than messy real-world data.
+3. **Some comorbidity/complication flags are marginal draws.** Drawn at fixed (age-adjusted where
+   relevant) prevalences rather than jointly conditioned on the full clinical picture.
+4. **Mortality is a calibrated Bernoulli.** `inpatient_mortality_flag` is drawn on a severity- and
+   type-derived probability; time-to-death is an exponential rather than a fully integrated survival
+   model. Small CURB-65=0 subsets can show noisy mortality at low n (converges to target at scale).
+
+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-RES-003 product |
+|---|---|---|
+| Patients | 500 | 20,000+ (configurable) |
+| Seeds / cohorts | 1 | Multi-seed, reproducible |
+| Formats | CSV | CSV + Parquet + JSON |
+| Comorbidity model | Marginal draws | Jointly conditioned |
+| Mortality | Cross-sectional Bernoulli | Integrated survival model |
+| License | CC-BY-NC-4.0 | Commercial |
+| Support & SLA | — | Included |
+
+Full product, custom cohorts, or other respiratory SKUs (asthma, COPD, ILD, pulmonary hypertension,
+sleep apnea): **pradeep@xpertsystems.ai**
+
+---
+
+## Citation
+
+```bibtex
+@dataset{xpertsystems_hc_res_003_2026,
+  title        = {HC-RES-003: Pneumonia 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: ATS/IDSA 2019 CAP
+                  guidelines; PSI/Fine score (Fine et al. 1997, NEJM); CURB-65 (Lim et al.
+                  2003, Thorax); SMART-COP (Charles et al. 2008, CID); ARDSNET (NEJM 2000)
+                  and Berlin ARDS definition; RECOVERY trial (Horby et al. 2021, NEJM); CDC
+                  NHSN HAP/VAP surveillance; CMS HRRP readmission benchmarks; Surviving Sepsis
+                  Campaign 2021. Third SKU in the XpertSystems Respiratory vertical.}
+}
+```
+
+*Synthetic data generated by XpertSystems.ai. Not derived from real patient records.
+Not for clinical use.*