Tobacco as a risk factor of larynx cancer: Artificial neural network modeling


Tags : tobacco, larynxcancer, riskfactors, artificialintelligence, artificialneuralnetworks

Category : Original articles

Authors : Bouaoud Souad, Bouharati Khaoula, Kara Lamia, Ziadi Zoubida, Boukharouba Hafida, Bouharati Saddek, Mahnane Abbas, Hamdi-cherif Mokhtar

Full text (PDF file)

Background: Several factors are involved in the development of laryngeal cancer. Tobacco is considered a major factor. However, the effect of these factors and the weight of each vary from one person to another. As far as human physiological functions are concerned, the processes are very complex. The analysis of these data remains very complex and very difficult if it is impossible to process them using the classical mathematical tool. This study proposes the application of a tool based on the principles of artificial intelligence including artificial neural networks. Material and Methods:This study included several factors are involved in the development of laryngeal cancer. Tobacco is considered a major factor. From the cases recorded in the district of Setif in Algeria during a decade. An artificial neural network model is proposed in the data analysis. Results: the proposed artificial neural network system can bypass all complexity and constitute a prevention tool. The constructed system consists of seven layers of input layer (Age, year, gender, crude rate, ASR(WR), Cum incidence, % on total cancers and median age), hidden layer and an output layer expressing the number of cases recorded with laryngeal cancer. Conclusion: The parameters related to the recorded subjects are considered as input factors to the system the number of cases recorded is considered as output variable. The system establishes a function of correspondence between the inputs and the output from the learning phase of the network. This makes it possible to introduce random variables at the inputs of the system to read instantly and with great precision the probable number of cases at the output. This tool can be a means of preventing this type of cancer. The system is able to manage all the complexity related to the phenomenon and therefore to predict the appearance of this type of cancer.