一种连续性肾脏替代治疗设备故障预测方法及系统
By constructing a multi-node association graph and using a support vector machine classification model and a long short-term memory network model, the correlation pattern between the increase in transmembrane pressure and the decrease in net ultrafiltration rate of CRRT equipment is identified, and a fault probability distribution is generated. This solves the problem of low sensitivity in CRRT equipment fault prediction and improves the scientificity and safety of equipment condition assessment.
Patent Information
- Authority / Receiving Office
- CN · China
- Patent Type
- Patents(China)
- Current Assignee / Owner
- THE FIRST AFFILIATED HOSPITAL OF ARMY MEDICAL UNIV
- Filing Date
- 2026-04-02
- Publication Date
- 2026-07-17
AI Technical Summary
In the existing technology, the fault prediction sensitivity of continuous renal replacement therapy (CRRT) devices is low, and it is unable to accurately capture early signals of device performance degradation. This results in insufficient sensitivity and reliability of risk identification when dynamic risks such as filter blockage occur, increasing the probability of unplanned treatment interruption.
A graph neural network is used to construct a multi-node association graph. Combined with a support vector machine classification model and a long short-term memory network model, the association pattern between increased transmembrane pressure and decreased net ultrafiltration rate is identified through multi-factor linkage association path features. Combined with Bayesian inference to generate a failure probability distribution, automatic switching operation is realized to reduce the probability of unplanned interruption of the treatment process.
It significantly improves the scientific rigor and safety of equipment condition assessment, enabling more sensitive capture of subtle dynamic correlations between parameters, reducing reliance on fixed alarm thresholds, enhancing the accuracy of identifying blockage trend signals, and ensuring the equipment's autonomous and continuous operation capability under high-risk conditions.
Smart Images
Figure CN122050756B_ABST