UDF-GMA: Uncertainty Disentanglement and Fusion for General Movement Assessment

TL;DR

UDF-GMA introduces a novel approach to automated general movement assessment by explicitly modeling and fusing epistemic and aleatoric uncertainties, improving reliability and generalizability in clinical settings.

Contribution

It is the first to disentangle and fuse epistemic and aleatoric uncertainties in pose-based GMA, enhancing prediction robustness.

Findings

Effective uncertainty disentanglement improves classification accuracy.

Fusion of uncertainties with motion features enhances class separation.

Demonstrated superior performance on Pmi-GMA benchmark dataset.

Abstract

General movement assessment (GMA) is a non-invasive tool for the early detection of brain dysfunction through the qualitative assessment of general movements, and the development of automated methods can broaden its application. However, mainstream pose-based automated GMA methods are prone to uncertainty due to limited high-quality data and noisy pose estimation, hindering clinical reliability without reliable uncertainty measures. In this work, we introduce UDF-GMA which explicitly models epistemic uncertainty in model parameters and aleatoric uncertainty from data noise for pose-based automated GMA. UDF-GMA effectively disentangles uncertainties by directly modelling aleatoric uncertainty and estimating epistemic uncertainty through Bayesian approximation. We further propose fusing these uncertainties with the embedded motion representation to enhance class separation. Extensive experiments on the Pmi-GMA benchmark dataset demonstrate the effectiveness and generalisability of the proposed approach in predicting poor repertoire.