Evaluating Latent Space Robustness and Uncertainty of EEG-ML Models under Realistic Distribution Shifts
This work addresses the challenge of ensuring reliable EEG-based machine learning models for healthcare applications, though it is incremental as it extends existing evaluations with new diagnostic analyses.
The study tackled the problem of limited clinical utility of EEG-ML models under real-world data shifts by developing diagnostic measures to detect performance degradation before deployment, finding that latent space integrity and model uncertainty under data transforms can help anticipate such issues.
The recent availability of large datasets in bio-medicine has inspired the development of representation learning methods for multiple healthcare applications. Despite advances in predictive performance, the clinical utility of such methods is limited when exposed to real-world data. This study develops model diagnostic measures to detect potential pitfalls before deployment without assuming access to external data. Specifically, we focus on modeling realistic data shifts in electrophysiological signals (EEGs) via data transforms and extend the conventional task-based evaluations with analyses of a) the model's latent space and b) predictive uncertainty under these transforms. We conduct experiments on multiple EEG feature encoders and two clinically relevant downstream tasks using publicly available large-scale clinical EEGs. Within this experimental setting, our results suggest that measures of latent space integrity and model uncertainty under the proposed data shifts may help anticipate performance degradation during deployment.