Early risk prediction method based on chaotic feature fusion of multi-modal physiological signals

By constructing the chaotic attractor geometry of multimodal physiological signals and the cross-modal critical state correlation map, the problem that existing technologies cannot effectively characterize the evolution of the pathological state of heart disease is solved, and high-precision prediction and personalized intervention for early risk of heart disease are achieved.

CN122440134APending Publication Date: 2026-07-24钰兔科技集团有限公司
View PDF 0 Cites 0 Cited by

Patent Information

Authority / Receiving Office
CN · China
Patent Type
Applications(China)
Current Assignee / Owner
钰兔科技集团有限公司
Filing Date
2026-06-03
Publication Date
2026-07-24

AI Technical Summary

Technical Problem

Existing methods for early risk prediction of heart disease cannot effectively characterize the slow drift process of physiological systems from a steady state to an unstable state, cannot identify heterocentric orbital convergence patterns between multimodal attractors, and have insufficient model adaptability, resulting in limited prediction accuracy and inaccurate results.

Method used

By collecting time-series data from multiple heterogeneous physiological signal sources, a chaotic attractor geometry of multimodal physiological signals is constructed, the distribution of invariant manifolds and saddle points is identified, heteroclinic orbital convergence patterns are tracked, a cross-modal critical state correlation map is constructed, the potential energy function distribution is calculated, and real-time correction is performed using the Lyapunov exponent spectrum to form a dynamically self-consistent closed loop.

Benefits of technology

It significantly improves the ability to analyze precursors of high-risk states, reduces the false alarm rate, and enables continuous quantitative assessment of the direction and intensity of physiological state evolution towards risk, thereby enhancing the stability and reliability of the method for personalized early intervention in heart disease.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN122440134A_ABST
    Figure CN122440134A_ABST
Patent Text Reader

Abstract

The present application relates to the technical field of physiological signal processing, and particularly relates to an early risk prediction method based on multi-modal physiological signal chaotic feature fusion. Through acquisition of multi-modal physiological signal time series data and phase space reconstruction to form a chaotic attractor, identification of an invariant manifold and a saddle point and tracking of a heteroclinic orbit to construct a cross-modal critical state correlation atlas, calculation of a geodesic distance between a current physiological state point and a critical transition path based on the atlas to generate a potential energy function distribution, deduction of an optimal evolution trajectory and a bifurcation point position under multi-modal collaborative constraints to generate an early risk evaluation result, identification of orbit drift using an actual observation sequence and backtracking of a Lyapunov exponent spectrum to rescale the invariant manifold topological dimension and the saddle point position to synchronously correct the correlation atlas to form a dynamically self-consistent closed loop. The method can accurately predict early cardiac risk and realize dynamic self-consistency, thereby improving evaluation accuracy.
Need to check novelty before this filing date? Find Prior Art