Developmental language disorder early identification method and system fusing neural activity features

By employing differential normalization and mutual information attention mechanisms, combined with brain network modulation matrices and clinical factor modulation, a composite model was constructed. This model addresses the problem of existing technologies failing to capture characteristic physiological differences and nonlinear dependencies, enabling the early identification of developmental language disorders.

CN122050876BActive Publication Date: 2026-07-03BEIHANG UNIV
View PDF 2 Cites 0 Cited by

Patent Information

Authority / Receiving Office
CN · China
Patent Type
Patents(China)
Current Assignee / Owner
BEIHANG UNIV
Filing Date
2026-04-16
Publication Date
2026-07-03

AI Technical Summary

Technical Problem

In the early identification of developmental language disorders, conventional normalization methods ignore the physiological differences in power features and coherence features, cannot effectively quantify the nonlinear dependency between features and category labels, and lack the fusion of prior knowledge about brain networks. This results in the model's insufficient ability to model individual differences and the hierarchical structure of brain networks, making it difficult to explain the model's decision-making process.

Method used

A differentiated normalization strategy is adopted to target power features and coherence features. Combined with brain network modulation matrix and mutual information attention mechanism, feature importance is dynamically evaluated. The feature vector is mapped by radial basis function to construct a composite model that integrates brain region-level feature aggregation and clinical factor modulation. The model is optimized using binary cross-entropy loss function.

Benefits of technology

It enhances feature discrimination, improves the model's nonlinear expressive power and neurophysiological interpretability, and can more accurately identify developmental language disorders, capturing the complex interaction between local brain activity and network connectivity information.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN122050876B_ABST
    Figure CN122050876B_ABST
Patent Text Reader

Abstract

The application relates to a development language disorder early identification method and system fusing neural activity features and belongs to the technical field of artificial intelligence. The method comprises the following steps: acquiring original neurophysiological signals and matched clinical information, and constructing a data set; power features and coherence features of the signals are normalized to generate enhanced feature vectors; attention weighted feature vectors are obtained through mutual information based attention mechanism weighting; the radial basis function is used to nonlinearly map the feature vectors to a high-dimensional space to obtain extended feature vectors; a development language disorder early identification model is constructed, brain region level feature aggregation and modulation, clinical factor fusion and gate modulation are adopted, a decision function fusing nonlinear interaction items is introduced, and a binary cross entropy loss function is used to optimize the model; after model training is completed by using a training set, signals of a subject to be evaluated are input into the model after the same preprocessing, and an identification result is output. The application can improve the accuracy of development language disorder early identification.
Need to check novelty before this filing date? Find Prior Art