Mining biomarkers from routine laboratory tests in clinical records associated with air pollution health risk assessment

• Published in: Environmental research Vol. 216; no. Pt 3; p. 114639
• Main Authors: Deng, Zhonghua, Tan, Chaochao, Pan, Jianhua, Xiang, Yangen, Shi, Guomin, Huang, Yue, Xiong, Yican, Xu, Keqian
• Format: Journal Article
• Language: English
• Published: Netherlands Elsevier Inc 01.01.2023
• Subjects: air; Air pollutants; Air Pollutants - analysis; Air Pollutants - toxicity; air pollution; Air Pollution - analysis; biomarkers; Biomarkers - analysis; China; Cholesterol, LDL; Clinical laboratory test; Clinical record; Data mining; health effects assessments; Health risk assessment; hospitals; Nitrogen Dioxide - analysis; Particulate Matter - analysis; Risk Assessment; statistical models; urea; Urea - analysis; China; Clinical record; Air pollutants; Data mining; Health risk assessment; Clinical laboratory test
• ISSN: 0013-9351; 1096-0953; 1096-0953
• DOI: 10.1016/j.envres.2022.114639

Clinical laboratory in hospital can produce amounts of health data every day. The purpose of this study was to mine biomarkers from clinical laboratory big data associated with the air pollution health risk assessment using clinical records. 13, 045, 629 clinical records of all 27 routine laboratory tests in Changsha Central Hospital, including ALB, TBIL, ALT, DBIL, AST, TP, UREA, UA, CREA, GLU, CK, CKMB, LDL-C, TG, TC, HDL-C, CRP, WBC, Na, K, Ca, Cl, APTT, PT, FIB, TT, RBC and those daily air pollutants concentration monitoring data of Changsha, including PM2.5, PM10, SO2, NO2, CO, and O3 from 2014 to 2016, were retrieved. The moving average method was used to the biological reference interval was established. The tests results were converted into daily abnormal rate. After data cleaning, GAM statistical model construction and data analysis, a concentration-response relationship between air pollutants and daily abnormal rate of routine laboratory tests was observed. Our study found that PM2.5 had a stable association with TP (lag07), ALB (lag07), ALT (lag07), AST (lag07), TBIL (lag07), DBIL (lag07), UREA (lag07), CREA (lag07), UA (lag07), CK (lag 06), GLU (lag07), WBC (lag07), Cl (lag07) and Ca (lag07), (P < 0.05); O3 had a stable association with AST (lag01), CKMB (lag06), TG (lag07), TC (lag05), HDL-C (lag07), K (lag05) and RBC (lag07) (P < 0.05); CO had a stable association with UREA (lag07), Na (lag7) and PT (lag07) (P < 0.05); SO2 had a stable association with TP (lag07) and LDL-C (lag0) (P < 0.05); NO2 had a stable association with APTT (lag7) (P < 0.05). These results showed that different air pollutants affected different routine laboratory tests and presented different pedigrees. Therefore, biomarkers mined from routine laboratory tests may potentially be used to low-cost assess the health risks associated with air pollutants. •Clinical laboratory in hospital produces amounts of health data every day.•Data mining from over 13 million clinical records in 27 routine laboratory tests.•Some clinical laboratory tests can be biomarkers of air pollution.•Quantitative relationship exists between air pollutants and mined biomarkers.•Clinical records can be a low-cost source of biomarkers with air pollution.