ECG-based Identity Validation during Bathing in Different Water Temperature^*.

Area of Science:

• Biometrics
• Cardiovascular Signal Processing
• Human-Computer Interaction

Background:

• Identity validation is crucial for security.
• Existing biometric methods can be invasive or inconvenient.
• Electrocardiogram (ECG) signals offer a potential non-invasive biometric modality.

Purpose of the Study:

• To propose and evaluate a novel identity validation method using ECG signals.
• To investigate the impact of bathing water temperature on ECG-based identity validation.
• To develop a robust classification model for accurate identity verification.

Main Methods:

• Collected ECG data from 10 subjects across 5 different bathtub water temperatures (37-41°C).
• Processed ECG signals including filtering, R-peak detection, and QRS complex segmentation.
• Utilized a long short-term memory (LSTM) network for identity classification, employing cross-validation.

Main Results:

• Identified that bathtub water temperature significantly influences identity validation accuracy.
• Achieved a highest identity validation accuracy of 98.43%.
• Attained an average identity validation accuracy of 97.68% using a model trained on data from all tested temperatures.

Conclusions:

• ECG signals acquired during bathing can be effectively used for identity validation.
• A unified LSTM model trained across various water temperatures provides high accuracy.
• This method offers a promising non-invasive biometric solution adaptable to different environmental conditions.