Development of data classification using a hybrid method of adaptiveartificial neural networks and particle swarm optimizationfor identifyingpatients at risk of diabetes
Abstract:
This research aimed to 1) develop a method of data classification using Adaptive Artificial Neural Network and Particle Swarm Optimization (AANN-PSO),
2) compare the performance of the developed data classification method with three types: Adaptive Artificial Neural Network and Particle Swarm Optimization (AANN-PSO), Artificial Neural Network and Particle Swarm Optimization (ANN-PSO) and Artificial Neural Network (ANN) and 3), classify the patients who are at risk of diabetes and normal subjects with the method of Adaptive Artificial Neural Network and Particle Swarm Optimization (AANN-PSO). The data set involved 7,000 patients who were at risk of diabetes, in the area under the responsibility of the Nakhon Phanom Provincial Health Office in the year 2018. The research results were as follows:
1. The data classification using Adaptive Artificial Neural Network and Particle Swarm Optimization (AANN-PSO) with the new conversion function when acted to decrease the slope of the target function, while data classification performance increased.
2. Data classification using AANN-PSO resulted in better performance than ANN-PSO and ANN in all five situations. Furthermore, when the sample size was increased, the performance was even better.
3. Factors that affected the risk of diabetes included body mass index, diastolic blood pressure, age, systolic blood pressure, and a family history of diabetes. The classification of patients who are at risk of diabetes by using AANN-PSO had an accuracy of 92.79%, with mean square error of 0.07.