No Imputation Needed: A Switch Approach to Irregularly Sampled Time Series
This addresses the issue of bias and sub-optimal performance in time series analysis for domains like healthcare, offering a novel approach without imputation.
The paper tackles the problem of modeling irregularly-sampled time series by introducing SLAN, a method that avoids imputation and dynamically adapts using switches, achieving improved performance on datasets like MIMIC-III and Physionet 2012.
Modeling irregularly-sampled time series (ISTS) is challenging because of missing values. Most existing methods focus on handling ISTS by converting irregularly sampled data into regularly sampled data via imputation. These models assume an underlying missing mechanism, which may lead to unwanted bias and sub-optimal performance. We present SLAN (Switch LSTM Aggregate Network), which utilizes a group of LSTMs to model ISTS without imputation, eliminating the assumption of any underlying process. It dynamically adapts its architecture on the fly based on the measured sensors using switches. SLAN exploits the irregularity information to explicitly capture each sensor's local summary and maintains a global summary state throughout the observational period. We demonstrate the efficacy of SLAN on two public datasets, namely, MIMIC-III, and Physionet 2012.