Assessment of triglyceride and cholesterol in overweight people based on multiple linear regression and artificial intelligence model

• Published in: Lipids in health and disease Vol. 16; no. 1; p. 42
• Main Authors: Ma, Jing, Yu, Jiong, Hao, Guangshu, Wang, Dan, Sun, Yanni, Lu, Jianxin, Cao, Hongcui, Lin, Feiyan
• Format: Journal Article
• Language: English
• Published: London BioMed Central 20.02.2017; BioMed Central Ltd; Springer Nature B.V
• Subjects: Adult; Artificial intelligence; Biomedical and Life Sciences; Case-Control Studies; Cholesterol; Cholesterol - blood; Clinical Nutrition; Fasting; Health aspects; Humans; Hyperlipidemia; Life Sciences; Linear Models; Lipidology; Medical Biochemistry; Middle Aged; Models, Biological; Neural Networks (Computer); Overweight - blood; Overweight persons; Risk factors; Triglycerides; Triglycerides - blood; China; Regression; Back propagation artificial neural network; Triglyceride; Cholesterol; Overweight
• ISSN: 1476-511X; 1476-511X
• DOI: 10.1186/s12944-017-0434-5

Background The prevalence of high hyperlipemia is increasing around the world. Our aims are to analyze the relationship of triglyceride (TG) and cholesterol (TC) with indexes of liver function and kidney function, and to develop a prediction model of TG, TC in overweight people. Methods A total of 302 adult healthy subjects and 273 overweight subjects were enrolled in this study. The levels of fasting indexes of TG (fs-TG), TC (fs-TC), blood glucose, liver function, and kidney function were measured and analyzed by correlation analysis and multiple linear regression (MRL). The back propagation artificial neural network (BP-ANN) was applied to develop prediction models of fs-TG and fs-TC. Results The results showed there was significant difference in biochemical indexes between healthy people and overweight people. The correlation analysis showed fs-TG was related to weight, height, blood glucose, and indexes of liver and kidney function; while fs-TC was correlated with age, indexes of liver function ( P  < 0.01). The MRL analysis indicated regression equations of fs-TG and fs-TC both had statistic significant ( P  < 0.01) when included independent indexes. The BP-ANN model of fs-TG reached training goal at 59 epoch, while fs-TC model achieved high prediction accuracy after training 1000 epoch. Conclusions In conclusions, there was high relationship of fs-TG and fs-TC with weight, height, age, blood glucose, indexes of liver function and kidney function. Based on related variables, the indexes of fs-TG and fs-TC can be predicted by BP-ANN models in overweight people.