Integrating machine learning and traditional methods for cadmium prediction and bioavailability assessment in Paeoniae Radix Alba: a case study from Bozhou, Anhui Province.

Journal: Environmental geochemistry and health
Published Date:

Abstract

Soil heavy metal contamination, particularly cadmium (Cd), poses a significant risk to ecosystems and human health. This study investigates the distribution and bioavailability of Cd in soil and Paeoniae Radix Alba system from Qiaocheng District, Bozhou City, China. To measure and predict Cd concentrations, we employed traditional laboratory methods and machine learning models, such as the Backpropagation Neural Network (BPNN) and Convolutional Neural Network combined with Long Short-Term Memory units (CNN-LSTM). The results revealed that although Cd concentrations in the soil of the Bozhou area did not exceed standard limits, there was still evidence of enrichment. The concentration of Cd in Paeoniae Radix Alba is much lower than the national standard, and the low bioavailability shows high safety. BPNN-CP and CNN-LSTM-CP were both capable of predicting Cd in Paeoniae Radix Alba, and the CNN-LSTM-CP model showed high accuracy in predicting Cd concentration, with an R as high as 0.97. This study provides critical data supporting the safety assessment of Paeoniae Radix Alba from the Bozhou area and introduces a novel technical approach for predicting heavy metal pollution in Chinese herbal medicines.

Authors

  • Fang He
    Department of Applied Mathematics, The Hong Kong Polytechnic University, Hong Kong.
  • Quan Tang
    School of Computer Science, China West Normal University, Nanchong, 637009, Sichuan, China.
  • Dong Li
    Department of Cardiovascular Medicine, Lanzhou University Second Hospital, 730030 Lanzhou, Gansu, China.
  • Lingling Wang
    Department of Neurology, Shanghai Pudong New Area People's Hospital, Shanghai, China.
  • Guijian Liu
    School of Earth and Space Sciences, University of Science and Technology of China, Hefei, 230026, China.