Construction and Analysis of a Joint Diagnosis Model of Random Forest and Artificial Neural Network for Obesity

Jian Yu; Xiaoyan Xie; Yun Zhang; Feng Jiang; Chuyan Wu

doi:10.3389/fmed.2022.906001

Construction and Analysis of a Joint Diagnosis Model of Random Forest and Artificial Neural Network for Obesity

Front Med (Lausanne). 2022 May 23:9:906001. doi: 10.3389/fmed.2022.906001. eCollection 2022.

Authors

Jian Yu¹, Xiaoyan Xie¹, Yun Zhang¹, Feng Jiang², Chuyan Wu¹

Affiliations

¹ Department of Rehabilitation Medicine, The First Affiliated Hospital of Nanjing Medical University, Nanjing, China.
² Department of Neonatology, Obstetrics and Gynecology Hospital of Fudan University, Shanghai, China.

Abstract

Obesity is a significant global health concern since it is connected to a higher risk of several chronic diseases. As a consequence, obesity may be described as a condition that reduces human life expectancy and significantly impacts life quality. Because traditional obesity diagnosis procedures have several flaws, it is vital to design new diagnostic models to enhance current methods. More obesity-related markers have been discovered in recent years as a result of improvements and enhancements in gene sequencing technology. Using current gene expression profiles from the Gene Expression Omnibus (GEO) collection, we identified differentially expressed genes (DEGs) associated with obesity and found 12 important genes (CRLS1, ANG, ALPK3, ADSSL1, ABCC1, HLF, AZGP1, TSC22D3, F2R, FXN, PEMT, and SPTAN1) using a random forest classifier. ALPK3, HLF, FXN, and SPTAN1 are the only genes that have never been linked to obesity. We also used an artificial neural network to build a novel obesity diagnosis model and tested its diagnostic effectiveness using public datasets.

Keywords: artificial neural network; diagnosis model; gene sequencing technology; obesity; random forest classifier.