Page Header

The Comparison of Classification Efficiency of Patients Who are Risk Diabetes

Akanit Thongjit, Poonpong Suksawang, Jatupat Mekparyup


This research aimed to compare efficiency model for classify patients who are risk diabetes. There are two data classification method used, Back Propagation Neural Networks (BPNN) and Back Propagation Neural Networks with Particle Swarm Optimization (BPNN-PSO). The data sets were provided by subjects who are risk to develop diabetes and are more than 35 years old, living in the area under the responsibility of Nakhon Phanom Provincial Health Office in year 2018. The total number of records was 7,000 cases. The data was analyzed by using technique on 10-Fold Cross Validation. Factors that affect the risk of diabetes included 10 variables: age, sex, smoke, alcohol, diabetes medical in family (DMFAMILY), hypertension in family (HTFAMILY), body mass index (BMI), waist, systolic blood pressure (SBP) and diastolic blood pressure (DBP). The results have shown that the BPNN-PSO model provided the accuracy in classification the diabetes higher than that of the BPNN model at 90.57% and the Mean Square Error lower than the BPNN models at MSE=0.09.


Artificial Neural Networks; Particle Swarm Optimization; Diabetes

Full Text: PDF


  • There are currently no refbacks.