Self-supervised Representation Learning on Electronic Health Records with Graph Kernel Infomax
This addresses the challenge of unsupervised representation learning on EHRs for clinical decision support, offering a novel method to handle complex temporality, though it appears incremental in the context of contrastive learning.
The paper tackles the problem of learning representations from Electronic Health Records (EHRs) for clinical applications like medication outcome prediction, proposing Graph Kernel Infomax, a self-supervised graph kernel learning approach that uses Kernel Subspace Augmentation to create manifold views without altering graph structures, and it empirically exceeds state-of-the-art performance on benchmark datasets.
Learning Electronic Health Records (EHRs) representation is a preeminent yet under-discovered research topic. It benefits various clinical decision support applications, e.g., medication outcome prediction or patient similarity search. Current approaches focus on task-specific label supervision on vectorized sequential EHR, which is not applicable to large-scale unsupervised scenarios. Recently, contrastive learning shows great success on self-supervised representation learning problems. However, complex temporality often degrades the performance. We propose Graph Kernel Infomax, a self-supervised graph kernel learning approach on the graphical representation of EHR, to overcome the previous problems. Unlike the state-of-the-art, we do not change the graph structure to construct augmented views. Instead, we use Kernel Subspace Augmentation to embed nodes into two geometrically different manifold views. The entire framework is trained by contrasting nodes and graph representations on those two manifold views through the commonly used contrastive objectives. Empirically, using publicly available benchmark EHR datasets, our approach yields performance on clinical downstream tasks that exceeds the state-of-the-art. Theoretically, the variation on distance metrics naturally creates different views as data augmentation without changing graph structures.