New Test-Time Scenario for Biosignal: Concept and Its Approach

TL;DR

This paper introduces a new test-time adaptation scenario for biosignal analysis, combining supervised and self-supervised learning to improve model robustness in real-time healthcare applications.

Contribution

It proposes a novel test-time scenario with mixed labeled and unlabeled biosignal data, and a framework that enhances model adaptability using dual-queue buffers and weighted sampling.

Findings

Improved accuracy in biosignal prediction tasks.

Enhanced model adaptability in real-world conditions.

Effective combination of supervised and self-supervised learning.

Abstract

Online Test-Time Adaptation (OTTA) enhances model robustness by updating pre-trained models with unlabeled data during testing. In healthcare, OTTA is vital for real-time tasks like predicting blood pressure from biosignals, which demand continuous adaptation. We introduce a new test-time scenario with streams of unlabeled samples and occasional labeled samples. Our framework combines supervised and self-supervised learning, employing a dual-queue buffer and weighted batch sampling to balance data types. Experiments show improved accuracy and adaptability under real-world conditions.