Research on Adverse Drug Reaction Prediction Model Combining Knowledge Graph Embedding and Deep Learning

Yufeng Li,Wenchao Zhao,Bo Dang,Xu Yan,Weimin Wang,Min Gao,Mingxuan Xiao

2024-07-27

Abstract:In clinical treatment, identifying potential adverse reactions of drugs can help assist doctors in making medication decisions. In response to the problems in previous studies that features are high-dimensional and sparse, independent prediction models need to be constructed for each adverse reaction of drugs, and the prediction accuracy is low, this paper develops an adverse drug reaction prediction model based on knowledge graph embedding and deep learning, which can predict experimental results. Unified prediction of adverse drug reactions covered. Knowledge graph embedding technology can fuse the associated information between drugs and alleviate the shortcomings of high-dimensional sparsity in feature matrices, and the efficient training capabilities of deep learning can improve the prediction accuracy of the model. This article builds an adverse drug reaction knowledge graph based on drug feature data; by analyzing the embedding effect of the knowledge graph under different embedding strategies, the best embedding strategy is selected to obtain sample vectors; and then a convolutional neural network model is constructed to predict adverse reactions. The results show that under the DistMult embedding model and 400-dimensional embedding strategy, the convolutional neural network model has the best prediction effect; the average accuracy, F_1 score, recall rate and area under the curve of repeated experiments are better than the methods reported in the literature. The obtained prediction model has good prediction accuracy and stability, and can provide an effective reference for later safe medication guidance.

Quantitative Methods,Artificial Intelligence,Machine Learning

What problem does this paper attempt to address?

The paper aims to address several key issues in the prediction of adverse drug reactions:

1. **High-dimensional sparse features**: In previous studies, the high dimensionality and sparsity of drug feature matrices led to inefficient model training and low prediction accuracy.
2. **Independent prediction models**: Each adverse drug reaction requires the construction of an independent prediction model, increasing the workload.
3. **Low prediction accuracy**: The prediction accuracy of existing methods needs improvement.

To solve these problems, the paper proposes a model for predicting adverse drug reactions that combines knowledge graph embedding and deep learning. Specifically:

- **Knowledge graph embedding technology**: By embedding drugs and their related features (such as targets, indications, enzymes, and carrier proteins) into a continuous low-dimensional feature space, the model integrates the association information between drugs, alleviating the issue of high-dimensional sparse features.
- **Deep learning model**: Utilizing a Convolutional Neural Network (CNN) model for efficient training, the prediction accuracy is improved.

This model can predict multiple adverse drug reactions within a unified framework, offering high prediction accuracy and stability, thus providing an effective reference for clinical safe medication.