基于图隐式非线性扩散和相似性预测miRNA-疾病关联性方法、装置及设备
By constructing an integrated similarity matrix and nonlinear diffusion graph neural network model, combined with a multilayer perceptron, the problems of insufficient information utilization and lack of aggregation of neighbor node information in miRNA-disease association prediction were solved, achieving highly accurate prediction results.
CN116825369BActive Publication Date: 2026-07-17HUNAN UNIV OF CHINESE MEDICINE
Patent Information
- Authority / Receiving Office
- CN · China
- Patent Type
- Patents(China)
- Current Assignee / Owner
- HUNAN UNIV OF CHINESE MEDICINE
- Filing Date
- 2023-03-29
- Publication Date
- 2026-07-17
AI Technical Summary
Technical Problem
Existing miRNA-disease association prediction methods fail to fully utilize various similarity data and neighbor node information, resulting in low prediction accuracy.
Method used
An integrated similarity matrix between miRNAs and diseases was constructed. Embedded features were learned using a graph neural network model based on nonlinear diffusion, and correlations were predicted using a multilayer perceptron, combining multi-source information and neighbor node information.
Benefits of technology
It improved the accuracy of miRNA-disease association prediction, achieving an AUC of 92.74±0.05% and an AUPR of 92.26±0.03% in five-fold cross-validation, and confirmed the accuracy of the top 50 miRNAs in case validation.
✦ Generated by Eureka AI based on patent content.
Smart Images
Figure CN116825369B_ABST
Abstract
本发明公开了一种基于图隐式非线性扩散和相似性预测miRNA‑疾病关联性方法，包括如下步骤：构建miRNA‑疾病关联矩阵，利用miRNA和疾病的多源信息构建miRNA集成相似性矩阵和疾病集成相似性矩阵，并将miRNA集成相似性矩阵和疾病集成相似性矩阵分别与miRNA‑疾病关联矩阵拼接构成miRNA和疾病的初始特征矩阵；利用基于非线性扩散的图神经网络模型从miRNA‑疾病关联网络学习miRNA和疾病的嵌入特征；拼接miRNA和疾病的嵌入特征，并利用多层感知机预测miRNA和疾病之间的关联性。本发明提供的预测miRNA‑疾病关联性方法，能有效提高miRNA‑疾病关联预测的准确性。本发明还提供一种基于图隐式非线性扩散和相似性预测miRNA‑疾病关联性装置及设备。
Need to check novelty before this filing date? Find Prior Art