Prediction of the radon concentration in thermal waters using artificial neural networks

The present paper focuses on predicting radon concentrations in thermal waters from three thermal water physicochemical properties: pH, temperature, and electrical conductivity. To achieve this, an artificial neural network model and a multiple regression analysis model were created. While developing both models, the data of 109 radon measurements in thermal waters acquired from the literature were employed. When the experimental radon concentrations were compared to those predicted by both models, the artificial neural network model predicted radon concentrations that were substantially closer to the experimental values. A variety of performance measures were also computed for evaluating both models' prediction ability. The artificial neural network model outperformed based on the measures computed, demonstrating the applicability and accuracy of the model in radon concentration prediction in thermal waters. The study demonstrates that the developed artificial neural network model for this research may be used to predict the radon concentration in thermal waters using three thermal water physicochemical parameters.