Graph-Guided Network for Irregularly Sampled Multivariate Time Series

TL;DR

RAINDROP is a graph neural network designed to model irregularly sampled multivariate time series by learning sensor dependencies directly from data, improving classification accuracy in healthcare and activity datasets.

Contribution

It introduces a novel message passing operator and sensor graph learning approach for irregular time series, outperforming existing methods.

Findings

RAINDROP achieves up to 11.4% higher F1-score than state-of-the-art.

It effectively models time-varying sensor dependencies.

The method performs well even when sensors are missing or unobserved.

Abstract

In many domains, including healthcare, biology, and climate science, time series are irregularly sampled with varying time intervals between successive readouts and different subsets of variables (sensors) observed at different time points. Here, we introduce RAINDROP, a graph neural network that embeds irregularly sampled and multivariate time series while also learning the dynamics of sensors purely from observational data. RAINDROP represents every sample as a separate sensor graph and models time-varying dependencies between sensors with a novel message passing operator. It estimates the latent sensor graph structure and leverages the structure together with nearby observations to predict misaligned readouts. This model can be interpreted as a graph neural network that sends messages over graphs that are optimized for capturing time-varying dependencies among sensors. We use RAINDROP to classify time series and interpret temporal dynamics on three healthcare and human activity datasets. RAINDROP outperforms state-of-the-art methods by up to 11.4% (absolute F1-score points), including techniques that deal with irregular sampling using fixed discretization and set functions. RAINDROP shows superiority in diverse setups, including challenging leave-sensor-out settings.