Intelligent aspiration monitoring method and system

By using multi-source data fusion and semi-supervised deep learning models, the invasiveness and accuracy issues of existing aspiration monitoring methods have been resolved, enabling non-invasive, real-time aspiration monitoring and early warning, thus improving the monitoring effect for patients with chronic respiratory diseases.

CN122440136APending Publication Date: 2026-07-24TIANJIN YOUAI REHABILITATION MEDICAL EQUIP
View PDF 0 Cites 0 Cited by

Patent Information

Authority / Receiving Office
CN · China
Patent Type
Applications(China)
Current Assignee / Owner
TIANJIN YOUAI REHABILITATION MEDICAL EQUIP
Filing Date
2026-06-10
Publication Date
2026-07-24

AI Technical Summary

Technical Problem

Existing methods for detecting aspiration rely on invasive or semi-invasive techniques, which cannot achieve long-term dynamic monitoring. Furthermore, the accuracy of single biosignal recognition is low, making it difficult to detect latent aspiration, resulting in insufficient sensitivity and specificity in aspiration detection.

Method used

Employing multi-source data fusion technology, including surface biosignals, dynamic laryngoscopy video data from the pharynx, and structured clinical data, aspiration identification is performed using a semi-supervised deep learning model. By combining self-attention mechanisms and metric learning to optimize feature fusion, non-invasive monitoring and real-time early warning are achieved.

Benefits of technology

It improves the accuracy and reliability of aspiration detection, enables long-term dynamic monitoring, reduces technical barriers and medical costs, supports rapid early warning and risk management, and is suitable for daily monitoring of patients with chronic respiratory diseases.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN122440136A_ABST
    Figure CN122440136A_ABST
Patent Text Reader

Abstract

The application discloses an intelligent aspiration monitoring method and system, and relates to the field of medical monitoring. In view of the problems of traditional aspiration monitoring, such as dependence on professional operation and single signal source, the method synchronously collects body surface biological signals, pharyngeal cavity dynamic laryngoscope videos and structured clinical data, and constructs a data set through preprocessing, double-blind labeling and integration. A semi-supervised deep learning algorithm is used to fuse multi-source features, an aspiration recognition model is constructed based on a transformer architecture, real-time recognition and sound-light early warning are realized. The system corresponds to data acquisition, processing and integration, model construction, training and prediction units. The application improves the aspiration recognition accuracy and generalization ability, can capture implicit aspiration, supports long-term dynamic monitoring, reduces medical costs and gains time for clinical intervention.
Need to check novelty before this filing date? Find Prior Art