State Geological Institute of Dionyz Stur, Mlynská dolina 1, 817 04, Bratislava, Slovak Republic.
Faculty of Economics and Business, Paneuropean University, Tematínska 10, 851 05, Bratislava, Slovak Republic.
Environ Geochem Health. 2017 Dec;39(6):1513-1529. doi: 10.1007/s10653-017-9944-3. Epub 2017 Mar 28.
For the evaluation of various adverse health effects of chemical elements occurring in the environment on humans, the comparison and linking of geochemical data (chemical composition of groundwater, soils, and dusts) with data on health status of population (so-called health indicators) play a key role. Geochemical and health data are predominantly nonlinear, and the use of standard statistical methods can lead to wrong conclusions. For linking such data, we find appropriate the use method of artificial neural networks (ANNs) which enable to eliminate data inhomogeneity and also potential data errors. Through method of ANNs, we are able to determine the order of influence of chemical elements on health indicators as well as to define limit values for the influential elements at which the health status of population is the most favourable (i.e. the lowest mortality, the highest life expectancy). For determination of dependence between the groundwater contents of chemical elements and health indicators, we recommend to create 200 ANNs. In further calculations performed for identification of order of influence of chemical elements as well as definition of limit values, we propose to work with median or mean values from calculated 200 ANNs. The ANN represents an appropriate method to be used for environmental and health data analysis in medical geochemistry.
为了评估环境中存在的化学元素对人类各种不良健康影响，将地球化学数据（地下水、土壤和尘埃的化学成分）与人口健康状况数据（所谓的健康指标）进行比较和关联，这起着关键作用。地球化学和健康数据主要是非线性的，使用标准统计方法可能会得出错误的结论。为了关联此类数据，我们发现人工神经网络（ANNs）的使用方法是合适的，它可以消除数据的不均匀性和潜在的数据错误。通过 ANN 方法，我们能够确定化学元素对健康指标的影响顺序，并定义影响元素的限值，在这些限值下，人口的健康状况最佳（即死亡率最低，预期寿命最高）。为了确定地下水中化学元素含量与健康指标之间的依赖关系，我们建议创建 200 个 ANN。在进一步进行的计算中，为了确定化学元素影响顺序以及定义限值，我们建议使用从 200 个计算的 ANN 中计算得出的中位数或平均值。ANN 是一种用于医学地球化学中环境和健康数据分析的合适方法。
