Physiological Signal Identification Using Machine Learning Models
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing biometric identification systems face accuracy issues due to the potential for stolen, lost, or forged bio-signals, such as fingerprints or voices, which can compromise user identification.
Innovation Solution
A method and system for identifying users based on physiological signals, including electrocardiogram (ECG) or photoplethysmogram (PPG) signals, using machine learning models like neural networks to extract and compare identification features, ensuring accurate matching.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If traditional biometric signals (fingerprint, voice, vein) are used for identification, then the identification process is simple and fast, but the security and accuracy are compromised due to potential theft, loss, or forgery of these signals
Solution Approach 1:
The patent transforms physiological signals from their raw form into extracted features and then into mathematical models, changing the parameter representation from direct signal comparison to model-based feature comparison. This increases reliability by using more stable, forged-resistant features while managing complexity through automated extraction processes
Solution Approach 2:
The patent introduces machine learning models as intermediaries between the physiological signals and the identification decision. These models extract meaningful features and create mathematical representations that are more resistant to forgery, thereby improving reliability while the automation of this process helps manage system complexity
2Measurement precision
If physiological signals are processed through multiple machine learning models to extract identification features, then the identification accuracy improves, but the processing time and computational complexity increase
Solution Approach 1:
The patent performs preliminary feature extraction and model generation during a registration phase before actual identification is needed. By pre-processing physiological data to create mathematical models and extract stable features in advance, the system reduces processing time during actual identification while maintaining high accuracy through the use of pre-computed models
Solution Approach 2:
The patent divides the identification process into distinct segments: signal acquisition, feature extraction, model generation, and identification decision. This segmentation allows each step to be optimized independently, with machine learning models handling feature extraction and comparison, thereby improving measurement precision while managing overall processing time through efficient division of tasks
Data Source
AI summary
A method may include acquiring first physiological data relating to a first subject, extracting at least one first physiological feature from the first physiological data relating to the first subject, determining a first model relating to at least one first reference physiological feature, generating, based on the first model and the at least one first physiological feature, a second model, the second model relating to at least one second reference physiological feature corresponding to the second model, and determining, based on the second model and the at least one first physiological feature, at least one identification physiological feature relating to the first subject. In some embodiments, the at least one identification physiological feature may correspond to the at least one second reference physiological feature.