Neural network assessment of metal retention in the body of adolescent children, depending on the air and water-food routes of intake

Direct assessment of retention is extremely complex, both due to the many internal mechanisms that ensure the dynamics of the metal content, and due to the huge variety of organs, tissues, processes that ensure their redistribution, transport and accumulation. The content of metals in the accumulating body environment is an integral indicator that summarizes the multi-environment impact and takes into account all the ways in which metals enter the body. A complex neural network model of metal retention in the body was developed, including ensembles of neural network regression models that calculate the levels of metal retention, depending on the place of residence of adolescent children and intake not only with consumed drinking water, but also with inhaled air. As a result of the studies, a database of metal concentrations in the atmospheric air, consumed drinking water, blood and urine was formed, taking into account the physiological characteristics of the tested sensitive group of adolescent children, with targeted territorial reference of the analyzed samples. The simplified structure of the neural network regression model (reducing the number of inputs) gives sufficient accuracy, and the reduction of neural net works increases the adequacy of the models.