New Test-Time Scenario for Biosignal: Concept and Its Approach
This work addresses the need for robust online adaptation in healthcare applications like blood pressure prediction, though it appears incremental as it builds on existing OTTA methods.
The paper tackles the problem of adapting pre-trained models to real-time biosignal prediction by introducing a new test-time scenario with streams of unlabeled and occasional labeled samples, resulting in improved accuracy and adaptability under real-world conditions.
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.