Identification of people at risk of chronic diseases at an early stage are of great significance. With the help of machine learning technology, the complex hidden correlation between chronic disease risk factors can be mined, and disease early warning models can be established. However, the accuracy of machine learning algorithm is closely related to the parameters of the model, which also restricts the application of machine learning in chronic disease prediction. In view of the above problems, this paper designs a solution to automatically predict the risk of chronic diseases based on Auto-Sklearn, evaluates and compares the models trained by Sklearn and Auto-Sklearn, and the accuracy of the model trained by Automated Machine Learning (AutoML) reaches 89.6%, which verifies the high performance of AutoML model. Furthermore, visualization technology is used to display the data cleaning process and explain the black box problem of AutoML model. This scheme demonstrates the usability in predicting chronic disease.
Research on Identification of Chronic Disease Using Automated Machine Learning
Lect. Notes Electrical Eng.
International Workshop of Advanced Manufacturing and Automation ; 2020 ; Zhanjiang, China October 12, 2020 - October 13, 2020
23.01.2021
9 pages
Aufsatz/Kapitel (Buch)
Elektronische Ressource
Englisch