Predicting the Associations between Meridians and Chinese Traditional Medicine Using a Cost-Sensitive Graph Convolutional Neural Network

Abstract

Natural products are the most important and commonly used in Traditional Chinese Medicine (TCM) for healthcare and disease prevention in East-Asia. Although the Meridian system of TCM was established several thousand years ago, the rationale of Meridian classification based on the ingredient compounds remains poorly understood. A core challenge for the traditional machine learning approaches for chemical activity prediction is to encode molecules into fixed length vectors but ignore the structural information of the chemical compound. Therefore, we apply a cost-sensitive graph convolutional neural network model to learn local and global topological features of chemical compounds, and discover the associations between TCM and their Meridians. In the experiments, we find that the performance of our approach with the area under the receiver operating characteristic curve (ROC-AUC) of 0.82 which is better than the traditional machine learning algorithm and also obtains 8%-13% improvement comparing with the state-of-the-art methods. We investigate the powerful ability of deep learning approach to learn the proper molecular descriptors for Meridian prediction and to provide novel insights into the complementary and alternative medicine of TCM.

Keywords: Meridian classification; Traditional Chinese Medicine; graph convolutional neural network.

Figure 1

The entire workflow of our study.

Figure 2

The simple illustration of the graph convolutional neural network (GCN) model.

Figure 3

Cosine similarity among Meridians.

Figure 4

(a) ROC-AUC performance among different numbers of hidden layers; (b) ROC-AUC performance among different numbers of hidden neurons.

Figure 5

The major substructure of the heart Meridian exists in the component of the herb Lily Bulb, Colchicine.

Figure 6

(a) Vascular smooth muscle cells were subjected to control medium (left upper) and high inorganic phosphate containing osteogenic medium without (middle upper) and with 0.1 (right upper), 1 (left lower), 5 (middle lower), and 10 (right lower) microM ATX. ATX, astaxanthin; Ctrl, control Pi, inorganic phosphate; ATX, astaxanthin; (b) the barplot of the relative alizarin red (AR) stain density.