Malignant Neoplasms in the Industrial City: Epidemiology, Current Trends and Forecast

Abstract

Introduction: Optimization of information and analytical support for public health monitoring based on modern techniques of mathematical modeling and forecasting, multivariate statistical methods and artificial neural networks is becoming highly relevant. Objective: To conduct a retrospective and prospective epidemiological analysis of the incidence of malignant neoplasms in the city of Taganrog, Rostov Region, for 1985–2022. Materials and methods: We did a comprehensive long-term data analysis of statistical reporting forms and a personalized database of malignant neoplasms using factor analysis, hierarchical cluster analysis, real risk assessment, and artificial neural networks. We used software of our own design, as well as software packages IBM SPSS Statistics version 19.0 and Matlab R2021a with the Neural Network Toolbox. Results: Our findings indicate an unfavorable situation in the city of Taganrog, Rostov Region, with a 1.3-fold excess of the average annual cancer rate for the cities of the region and a continuous rising trend. Sex and age characteristics and priority cancer sites were determined. Based on regional criteria for assessing the real risk, we established that Taganrog ranks first in terms of cancer incidence and mortality, as well as seven cancer sites. The highest real risk has been found for breast and skin cancer. Using factor analysis and hierarchical cluster analysis, we examined the structure of risk factors for colon cancer in 1988–2019. The applied technique of artificial neural networks provided higher accuracy in the medium-term forecasting of the frequency of malignant neoplasms compared to extrapolation forecasting using theoretical trend lines. Conclusion: The use of multivariate statistical methods and artificial neural networks provides a highly informative characterization of the health status of the population.