Method and system for reliability detection of prediction confidence of brain-computer interface device

By calculating the cumulative sum of individual calibration deviations, a reliability testing method for the predicted confidence of brain-computer interface devices is determined. By calculating the reliability assessment of the device, the reliability assessment of the predicted confidence of the system can be identified, which solves the problem of lack of reliability testing in the prior art and improves the security and interaction efficiency of the device.

CN122285456APending Publication Date: 2026-06-26BEIJING ZHONGGUANCUN SHUIMU MEDICAL TECH CO LTD
View PDF 0 Cites 0 Cited by

Patent Information

Application Number
CN202610450718.3
Authority / Receiving Office
CN · China
Patent Type
Applications(China)
Current Assignee / Owner
Filing Date
2026-04-08
Publication Date
2026-06-26

AI Technical Summary

Technical Problem

Existing brain-computer interface devices lack a detection scheme for the reliability of prediction confidence, which leads to the device performing dangerous operations or being overly conservative under incorrect assumptions, thus reducing interaction efficiency.

Method used

By calculating the sum of individual calibration biases for N intention-imagination tasks, the predictive confidence reliability of brain-computer interface devices is determined, including intention recognition, predictive confidence assessment, and calibration bias curve analysis, providing a quantifiable reliability assessment.

Benefits of technology

It enables quantitative assessment of the reliability of prediction confidence of brain-computer interface devices, can identify systematic error tendencies, eliminate interference from human subjective judgment, and more accurately assess whether the prediction confidence is reliable, thereby improving device security and interaction reliability.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN122285456A_ABST
    Figure CN122285456A_ABST
Patent Text Reader

Abstract

This invention provides a method and system for reliable detection of the prediction confidence of a brain-computer interface (BCI) device. The method includes: treating each intention-based visualization task as a sample; having the BCI device to be tested perform intention recognition on the current sample; obtaining the intention recognition result and prediction confidence output by the BCI device corresponding to the current sample; obtaining the actual error corresponding to the current sample based on the intention recognition result; obtaining the prediction error corresponding to the current sample based on the prediction confidence; obtaining the individual calibration deviation corresponding to the current sample based on the actual error and the prediction error; and determining the reliability of the prediction confidence of the BCI device to be tested by summing the individual calibration deviations of all samples. This invention can detect the reliability of the prediction confidence of a BCI device.
Need to check novelty before this filing date? Find Prior Art