Pain-related abnormal behavior ai recognition terminal for alzheimer's disease patients

CN122911474APending Publication Date: 2026-10-16GUANGZHOU VOCATIONAL & TECH COLLEGE OF HEALTH +1
0 Cites 0 Cited by

Patent Information

Application Number
CN202611045162.6
Authority / Receiving Office
CN · China
Patent Type
Applications(China)
Current Assignee / Owner
Filing Date
2026-07-14
Publication Date
2026-10-16

AI Technical Summary

Technical Problem

嵌入式终端运行过程中会存在资源占用波动、运算调度差异以及硬件负载变化，该类硬件工况变化会直接影响传感信号采集与运算的有效性，固定权重融合方式无法适配硬件工况变化带来的信号可信度差异，会导致无效传感数据参与特征融合运算

Benefits of technology

[0015] This invention addresses the core problems of existing technologies, such as poor adaptability to computing power fluctuations, low accuracy of fixed-weight fusion, and recognition deviations caused by latency interference. It achieves quantitative verification of signal validity by coupling the computing power load coefficient with the confidence level of the sensor signal, abandoning the traditional indiscriminate data processing mode. Simultaneously, it leverages signal confidence level to achieve multi-channel dynamic weight adaptive allocation, matching the signal characteristics of different sensor channels to complete differentiated nonlinear feature fusion. Combined with dual-mechanism latency correction to eliminate computational errors, this effectively solves the problems of inaccurate recognition of subtle pain behaviors in Alzheimer's patients and poor adaptability to various operating conditions, significantly improving the recognition stability and accuracy of the terminal under different hardware loads and computational latency scenarios.

✦ Generated by Eureka AI based on patent content.
Patent Text Reader

Abstract

The application belongs to the technical field of intelligent medical behavior recognition, and discloses an AI recognition terminal for pain-related abnormal behaviors of an Alzheimer's patient. The application comprises a multi-source sensing acquisition module, a signal normalization preprocessing module, a comprehensive confidence calculation module, a dynamic weight calculation module, a fusion feature calculation module, a time delay correction and criterion output module, an AI behavior recognition and communication storage module, and a computing power state monitoring module, wherein the computing power state monitoring module is electrically connected with the signal normalization preprocessing module. The application can adaptively distribute multi-channel feature weights by coupling the reliability of multi-source sensing signals with the terminal computing power working condition, and can complete pain behavior feature recognition by combining a differentiated nonlinear transformation and a time delay double mechanism correction, thereby solving the technical problems of traditional fixed weight recognition, non-adaptive computing power fluctuation and time delay interference, and effectively improving the adaptability and accuracy of weak pain abnormal behavior recognition of Alzheimer's patients.
Need to check novelty before this filing date? Find Prior Art