Evaluation of the applicability and clinical performance of carcinoembryonic antigen reference intervals using big laboratory data

Rigorous screening was performed according to the following criteria[3,4,9]: (1) exclusion of records with incomplete demographic information, such as sex or medical record number (n = 18); (2) for individuals with repeated measurements, retention of only the most recent CEA result, with 14,861 records excluded; (3) exclusion based on abnormal ultrasonographic findings, including biliary obstruction, cholestasis, hepatobiliary lithiasis, and liver cirrhosis (n = 18,952), thyroid nodules classified as TI-RADS category 3 or higher (n = 10,512), breast nodules classified as BI-RADS category 3 or higher (n = 1,725), and uterine fibroids or ovarian cysts (n = 1,235); abnormal chest CT findings, including ground-glass opacities, emphysema, pulmonary cysts, and space-occupying lesions (n = 16,524); and abnormal chest radiography or electrocardiography findings (n = 17,121); (4) systolic blood pressure ≥140 mmHg or diastolic blood pressure ≥90 mmHg (n = 7,408); (5) body mass index <18.5 kg/m² or ≥28 kg/m² (n = 6,296); and (6) abnormal laboratory results, including other common tumor markers, liver and renal function indices, metabolic indices, inflammatory markers, and common infectious-disease screening results outside the corresponding RIs (n = 55,995), to reduce confounding from pregnancy, acute infection, chronic disease, and occult malignancy.

The completeness rate for ancillary examinations, laboratory results, and personal health indices was 95.2%, yielding a final establishment dataset of 23,334 reference individuals.Evaluation datasetCEA measurements from patients attending FAHZU between November and December 2025 were stratified by clinical setting: outpatients (n = 52,081), health examinees (n = 29,674), and inpatients (n = 28,970).Disease datasetCEA measurements from patients with confirmed diagnoses at FAHZU between 2024 and 2025 were collected [16].