Dataset Overview
Definition：Million-Scale Health Checkup Datasetis provided byChangsha Langhui Information Technology Co., Ltd.Built on the DataAssetsAPI platform, this mega-structured medical database draws on more than 1 million reports from genuine in-depth examinations conducted by cooperative medical institutions and brand chains. Each report is carefully de-identified and graded through a five-dimensional quality rating system (S+/S/A/B/C).
| Dataset Name | Dataset marked in the 1 million-grade medical examination report |
| Total Data Volume | 1 million + medical examination reports, including vertical continuous data and high-end in-depth examinations |
| Time Span | 2019 - 2025 (5 years of vertical tracking) |
| Structured Fields | 800+ fields organized by L1-L4 project value ladder |
| Data Type | Comprehensive health examination report (PDF original + structured JSON) |
| Source Institution | Cooperative medical institutions and brand chains |
| Quality Grading | S+ S A B C Five-D scoring system |
| Percentage of abnormal cases | 80% for abnormal cases, avoiding homogenization of health data |
| Structure of the report | Four-part format: Basic information + description of what is seen + conclusions + doctor's recommendation/intervention |
| Image Integrity | All video type checks must be accompanied by original image screens (hard access threshold) |
| Format Output | PDF Original + Structured JSON (containing entity/relationship/value/unusual mark) |
Dataset schema: 800+ fields
The data set is layered by L1-L4 project value ladder, covering the full dimensions from basic medical examination to high-level precision medicine.
| Dimension Category | Field Count | Typical Fields | Value Tier |
|---|---|---|---|
| Basic demography | 12 | Exclusive mapping of ID, sex, year of birth, date of medical examination, age, BMI, blood pressure, height, weight, waist, hip | L1 Basic |
| Blood protocol. | 36 | Absolute values and percentages of white/red cell count, haemoglobin, red cell pressure, slab count, neutral/lymphoma/single/polythyx/alkaline cell count | L1 Basic |
| Pee General | 28 | Urea, pH, urine, urine, urin, urin, urin, blood, white cell esterase, urine microprotein, acetic anhydride, ACR | L1 Basic |
| General | 14 | Possibilities (immunization/chemics), penitent cells, penetrators, parasite eggs, trans-iron protein, calcium-backed proteins | L1 Basic |
| Biochemical. | 52 | Hepatic function (ALT/AST/GGT/ALP/TBIL/TP/ALB), kidney function (BUN/Cr/UA/CysC/eGFR), blood resin (TC/TG/HDL/LDL/ApoA1/ApoB), blood sugar (FPG/HbA1c/OGTT/insulin), electrolytics, myocardial enzymes, pancreas | L2 Health |
| tumor markers | 28 | AFP, CEA, CA19-9, CA125, CA15-3, CA72-4, PSA, fPSA, NSE, CYFRA21-1, SCC, ProGRP, HE4, CA242, Ferritin | L3 Advanced |
| Endocrine | 32 | TSH, FT3, FT4, TTab, TPOAb, Condensine, ACTH, GH, IGF-1, PRL, LH, FSH, E2, P, T, DHEA-S, Insulin/Cperium Release Test | L3 Advanced |
| Ultrasound Imaging | 86 | Abdominal ultrasound (papa/cholecal/pancin/spleen/trenal), thyroid ultrasound (TI-RADS), breast ultrasound (BI-RADS), cardiac ultrasound (IMT), EF/E/A, gynaecology/prostate ultrasound | L3 Advanced |
| CT / radiology | 48 | CT (Lung-RADS lung ectoplasmity), CTA (Agatston calcification fraction), DEXA (T/Z value), CT for the head, Cervical/Lowal MRI | L4 Premium |
| MRI/High-end Image | 42 | Head MRI/MRA (Fazekas classification, aneurysm), MRCP, breast MRI, prostate mpMRI (PI-RADS), PET-CT (SUVmax) | L4 Premium |
| Genetic testing | 36 | SEPTIN9 Mealification, Oncology Epidemiological Genomes (BrCA1/2, MMR Genetics, APC, TP53), genomics of pharmaceuticals (CYP2C19/CYP2D6/SLCO1B1/MTHFR) | L4 Premium |
| Endoscopy | 32 | Stomach mirrors (esophage/ gastrot/trestrophsic + pathology/HP/ atrophy/intestinalization), colon mirrors (salent meat/denoma, Paris/NICE spectrometry), bronchoscopys | L4 Premium |
| Specialized screening | 56 | HPV spectrometry, TCT (Bethesda), molybdenum target (BI-RADS), bone age (GP/TW3), lung function (FVC/FEV1/DLCO), arterial sclerosis (ABI/baPWV), HP exhalation test, sleep breathing monitoring (AHI), AI eye-floor screening | L3-L4 |
| Physical Examination | 28 | Medical (heart/pulmonary/birth/neurological), surgical (the thyroid/mistrogen/restalfinger) ophthalmology (sightedness/eye pressure/eye), ear, nose and throat, oral, gynaecology | L1 Basic |
| General review and recommendations | 18 | General examination conclusions (summary of abnormal findings), recommendations of the examining physician, risk hierarchy, recommended review cycle, recommended consultation room, time series analysis, health management recommendations | L3 Advanced |
| Metadata/Quality | 24 | Source level, reporting format, quality rating (S+/S/A/B/C), valid number of pages, number of video intercepts, de-identification, tagging of abnormal cases, vertical data ID, data year | — |
De-identified JSON Data Sample
The following is a structured JSON model of de-identified processing. The PII has been removed from the real name, ID number, mobile phone number, address, etc. and the core medical data is maintained. The exclusive map ID (anon_id) supports the vertical data series across the year.
{ "anon_id": "PHY-2023-A8F2K9M4X7", "report_id": "RPT-S-BJXH-20230415-0032", "meta": { "quality_grade": "S", "source_institution": "partner medical checkup centers", "institution_tier": "premium", "checkup_date": "2023-04-15", "report_pages": 42, "effective_pages": 38, "imaging_count": 8, "original_price_cny": 6800 }, "demographics": { "gender": "male", "birth_year": 1975, "age_at_checkup": 48, "height_cm": 172.5, "weight_kg": 78.3, "bmi": 26.3, "waist_cm": 92, "blood_pressure_mmhg": {"systolic": 138, "diastolic": 88} }, "lab_results": { "blood_routine": { "wbc_10e9_L": 6.8, "hgb_g_L": 152, "plt_10e9_L": 218 }, "biochemistry": { "alt_U_L": 42, "ggt_U_L": 68, "cr_umol_L": 88, "ua_umol_L": 420, "egfr_ml_min": 88, "tc_mmol_L": 5.82, "tg_mmol_L": 2.35, "ldl_c_mmol_L": 3.68, "fpg_mmol_L": 6.2, "hba1c_percent": 6.0 }, "tumor_markers": { "afp_ng_ml": 3.2, "cea_ng_ml": 2.1, "ca19_9_U_ml": 15.8, "psa_total_ng_ml": 2.8, "psa_ratio": 0.21 } }, "imaging": { "chest_ct": { "findings": "sample，sample，sample4mm，sample，Lung-RADS 2sample。", "lung_rads": "2", "nodule_count": 3, "largest_nodule_mm": 4 }, "carotid_us": { "findings": "sample，sample8.2×2.5mm，lumen stenosis ≈25%。", "imt_left_mm": 1.1, "stenosis_percent": 25 }, "cardiac_us": { "ef_percent": 62, "ea_ratio": 0.85 } }, "endoscopy": { "gastroscopy": { "findings": "sample，sample。sample：sample，HP(+)。", "hp_status": "sample value" } }, "summary": { "main_findings": [ "sample", "sample（sample25%）", "sample（IFG），HbA1cCNelevated", "mixed hyperlipidemia", "CNHPCN", "sample（Lung-RADS 2sample）" ], "risk_stratification": "sample value", "temporal_comparison": "sample2022sample：BMIelevated0.8kg/m²，CNelevated0.5mmol/L，sample" }
}AI Application Scenarios
Medical megamodel SFT training
1 million + complete four-part medical examination report + 800 + structured field to support fine-tuned alignment of the medical field of the GPT/Lama/Qwen model with RLHF.
Clinical decision support (CDSS)
S/S+ reports "Clinical chain of reasoning" data training CDSS model learning from a complete clinical thought chain from detection anomalies to diagnosis.
Health Risk Assessment Model
Five years of vertical continuity data build risk prediction models such as ASCVD, diabetes progress, early cancer screening, and time series matching fields provide key vertical signals.
Medical NLP / Information Extracting
Insight deduction level text supports NLP missions such as NER, relationship extraction, and medical text structure. 800+ fields are used as Gold Standard references.
Multimodal Medical AI
CT/MRI/PET-CT/ultrasound/endoscopy images paired with textual data and text reports, supporting multimodal vision-language model training and automatic image interpretation.
Pharmacogenomics and Personalized Medicine
Data such as CYP2C19/CYP2D6/SLCO1B1/MTHFR combined with medical examination indicators to train a personalized recommendation model for precision medication use.
Frequently asked questions
Need a custom medical dataset?
Changsha Langhui Information Technology Co., Ltd. The DataAssetsAPI platform is dedicated to providing AI enterprises and research institutions with high-quality, compliant medical data assets.
Data Sample Preview · Customized Data Solutions · Professional Technical Support