文章基本信息

标题：Construction of a genetic method to forecast the population health indicators based on neural network models
本地全文：下载
作者：Ievgen Fedorchenko ; Andrii Oliinyk ; Alexander Stepanenko 等
期刊名称：Eastern-European Journal of Enterprise Technologies
印刷版ISSN：1729-3774
电子版ISSN：1729-4061
出版年度：2020
卷号：1
期号：4
页码：52-63
DOI：10.15587/1729-4061.2020.197319
语种：English
出版社：PC Technology Center
摘要：A genetic method has been proposed to forecast the health indicators of population based on neural-network models. The fundamental difference of the proposed genetic method from existing analogs is the use of the diploid set of chromosomes in individuals in a population that is evolving. Such modification makes the dependence of the phenotype of the individual on the genotype less deterministic and, ultimately, helps preserve the diversity of the gene pool of the population and the variability of features of the phenotype during the execution of the algorithm. In addition, a modification of the genetic operator of mutations has been proposed. In addition, a modification genetic operator of mutations is proposed. In contrast to the classical method, those individuals that are exposed to the operator of mutations are selected not randomly but according to their mutation resistance corresponding to the value of the function of an individual adaptability. Thus, individuals with worse values of the target function are mutated, and the genome of the strong individuals remains unchanged. In this case, the likelihood of loss of the function reached during the evolution of the extremum due to the action of the mutation operator decreases, and the transition to the new extremum occurs if enough specific weight of the best attributes in the population is accumulated.A comparative analysis of the models synthesized with the help of the developed genetic method has shown that the best results were achieved in the model based on a neural network of long short-term memory. While creating and training the model based on a long short-term network, the ability to use the particle swarm method to optimize the network settings was investigated. The results of our experimental study have shown that the developed model yields the smallest error in predicting the number of new cases of tuberculosis – the average absolute error is 6.139, which is less compared with models that were built by using other methods).The practical application of the developed methods would make it possible to timely adjust the planned treatment and diagnostic, preventive measures, to determine in advance the necessary resources for localization and elimination of diseases in order to maintain people's health.
关键词：neural networks;genetic algorithm;phenotype;modified genetic mutation operator;forecasting of public health indicators