System and method for predicting oxygen desaturation

By analyzing the SpO2 signal of a pulse oximeter using a neural network, a predicted SpO2 value sequence is generated and the confidence level is compared. This solves the problem that pulse oximeters cannot predict oxygen desaturation in advance, enabling real-time prediction and timely alarms, and improving the reliability of prediction.

CN115334966BActive Publication Date: 2026-03-17COVIDIEN LP
View PDF 0 Cites 0 Cited by

Patent Information

Authority / Receiving Office
CN · China
Patent Type
Patents(China)
Current Assignee / Owner
Filing Date
2021-03-30
Publication Date
2026-03-17

AI Technical Summary

Technical Problem

Existing pulse oximeters can only indicate changes in oxygen saturation levels with a lag, and cannot predict oxygen desaturation in advance, making it impossible for clinicians to take timely preventive measures.

Method used

A neural network-based approach is employed to generate an input feature sequence of oxygen levels. The SpO2 signal is analyzed using a Long Short-Term Memory (LSTM) layer to generate a predicted SpO2 value sequence. The predicted values ​​are then compared with the input values ​​through a sequence-to-sequence prediction module to generate confidence values ​​and alarms, thereby enabling real-time prediction of impending oxygen level desaturation.

Benefits of technology

It enables real-time prediction of impending oxygen level desaturation, improves prediction confidence, and generates timely alerts to help clinicians take preventative measures.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN115334966B_ABST
    Figure CN115334966B_ABST
Patent Text Reader

Abstract

The detailed implementations described herein disclose a method of predicting oxygen desaturation, the method comprising: generating an input sequence of oxygen levels based on a sequence of input signals, the input signals being indicative of a physiological condition of a patient; generating an input feature sequence based on at least one of the sequence of input signals and the input sequence of oxygen levels; generating, using a neural network, a sequence of predicted values of the oxygen levels based on the input feature sequence; comparing the sequence of predicted values of the oxygen levels with the input sequence of oxygen levels over a predetermined time window to generate a sequence confidence value of the prediction; and generating an oxygen desaturation prediction based on the sequence of predicted values in response to determining that the sequence confidence value of the prediction is higher than a threshold confidence value.
Need to check novelty before this filing date? Find Prior Art