Ping An Makes Breakthrough in Artificial Intelligence-Driven Drug Research

Research by Ping An Healthcare Technology Research Institute and Tsinghua University has led to a promising deep learning framework for drug discovery, announced Ping An Insurance (Group) Company of China, Ltd. (hereafter “Ping An” or the “Group”, HKEX: 2318; SSE: 601318).

The findings were published in “An effective self-supervised framework for learning expressive molecular global representations to drug discovery” in Briefings in Bioinformatics, a peer-reviewed bioinformatics journal. It marks a major technology breakthrough for the Group in the field of AI-driven pharmaceutical research.

Drug discovery can take 10 to 15 years from invention to market. It can take a large number of experiments, with significant costs and high failure rates. Computer-aided drug discovery for molecule design in pre-clinical research helped to improve the process, but traditional methods were still expensive and time consuming. A variety of artificial intelligence technologies have shown superior speed and performance for different aspects of drug discovery, such as molecule drug design, drug-drug interaction and drug-target interaction predictions. However, molecular modeling has been a challenge, due to the limited amount of labelled data for training datasets.

Graph neural networks (GNN) have emerged as a powerful tool for modeling molecular data. Instead of relying on labelled data, a model can be pre-trained with unlabeled data. Ping An’s research proposed a novel deep learning framework, named MPG, that learns molecular representations from large volumes of unlabeled molecules, and a powerful GNN, called MolGNet, for modelling molecular graphs.

Ping An’s research also proposed a self-supervised pre-training strategy, named Pairwise Half-graph Discrimination, which was accepted in IJCAI 2021, a top peer-reviewed artificial intelligence conference. The research team found that after pre-training on 11 million unlabeled molecules, MolGNet can capture meaningful patterns of molecules to produce an interpretable representation. The experimental results showed that MPG achieved the state-of-the-art performance on multiple drug discovery tasks. It is an important first step towards graph-level self-supervised learning on large-scale molecule datasets.

The technology is being used by Ping An Shionogi, a joint venture founded by Ping An and Shionogi & Co., Ltd. a Japanese research-driven pharmaceutical company, for research and development of new drugs and drug repurposing.