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.

本文重点从三种热水物理化学特性（pH、温度和电导率）预测热水中的氡浓度。为了实现这一目标，创建了人工神经网络模型和多元回归分析模型。在开发这两个模型时，采用了从文献中获得的热水中 109 种氡气测量数据。当将实验氡气浓度与两个模型预测的氡气浓度进行比较时，人工神经网络模型预测的氡气浓度与实验值非常接近。还计算了各种性能指标来评估这两个模型的预测能力。根据计算的测量结果，人工神经网络模型表现出色，证明了该模型在温泉水中氡气浓度预测中的适用性和准确性。该研究表明，本研究开发的人工神经网络模型可用于使用三个热水物理化学参数来预测热水中的氡浓度。