IVP-VAE: Modeling EHR Time Series with Initial Value Problem Solvers
This work addresses computational bottlenecks in modeling EHR time series for healthcare applications, representing an incremental improvement over existing continuous-time models.
The paper tackled the computational inefficiency of continuous-time models for irregularly sampled EHR time series by proposing a method that models time series purely with continuous processes using an IVP solver, eliminating recurrent computation and enabling parallel state evolution. Experiments on three real-world datasets showed that the method outperforms predecessors, achieves state-of-the-art results, and offers significant advantages in data efficiency.
Continuous-time models such as Neural ODEs and Neural Flows have shown promising results in analyzing irregularly sampled time series frequently encountered in electronic health records. Based on these models, time series are typically processed with a hybrid of an initial value problem (IVP) solver and a recurrent neural network within the variational autoencoder architecture. Sequentially solving IVPs makes such models computationally less efficient. In this paper, we propose to model time series purely with continuous processes whose state evolution can be approximated directly by IVPs. This eliminates the need for recurrent computation and enables multiple states to evolve in parallel. We further fuse the encoder and decoder with one IVP solver utilizing its invertibility, which leads to fewer parameters and faster convergence. Experiments on three real-world datasets show that the proposed method can systematically outperform its predecessors, achieve state-of-the-art results, and have significant advantages in terms of data efficiency.