A method for disease multi-association prediction based on dynamic tensor graph learning
By employing a dynamic tensor graph learning method, a dynamic tensor graph structure is constructed. Combining multi-source heterogeneous data and time series information, this addresses the shortcomings of existing technologies in predicting the dynamic evolution of non-coding RNA associations with diseases and the insufficient modeling of heterogeneous nodes. It enables dynamic prediction of the synergistic regulatory relationship between miRNA-lncRNA-disease triples, thereby improving the accuracy of predictions and biological interpretability.
CN122245451APending Publication Date: 2026-06-19SHIHEZI UNIVERSITY
Patent Information
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- SHIHEZI UNIVERSITY
- Filing Date
- 2026-03-18
- Publication Date
- 2026-06-19
Smart Images
Figure CN122245451A_ABST
Abstract
This invention discloses a method for predicting multiple disease associations based on dynamic tensor graph learning, including step one, data acquisition and preprocessing; step two, constructing a dynamic tensor graph; step three, self-learning of the dynamic tensor graph; and step four, predictive analysis. This invention captures the dynamic evolution of miRNA, lncRNA, and disease associations through time window discretization and a time decay function, solving the problem that static models cannot reflect the temporal nature of biological processes. Furthermore, it combines the characteristic differences of miRNA, lncRNA, and disease, overcoming the limitations of single-type association modeling. Simultaneously handling three types of associations and their cascading effects, it provides a more comprehensive association map for the study of complex disease mechanisms. The time regularization term of the dynamic tensor decomposition and association evolution analysis clarify the key time points of association strength changes, providing biological clues for experimental verification.
Need to check novelty before this filing date? Find Prior Art