Patient Subtyping via Time-Aware LSTM Networks

TLDR

Patient heterogeneity leads to varied disease trajectories, and existing LSTM models assume regular time intervals, limiting their effectiveness on irregular longitudinal data. This study aims to develop a Time‑Aware LSTM (T‑LSTM) and a patient‑subtyping framework that can handle irregular time gaps in patient records. The T‑LSTM learns a subspace decomposition of cell memory to apply time‑decay discounting, and the model embeds patient sequences via an auto‑encoder for clustering into clinical subtypes. Experiments on synthetic and real datasets demonstrate that T‑LSTM captures underlying sequence structures with irregular time intervals, improving subtyping accuracy.

Abstract

In the study of various diseases, heterogeneity among patients usually leads to different progression patterns and may require different types of therapeutic intervention. Therefore, it is important to study patient subtyping, which is grouping of patients into disease characterizing subtypes. Subtyping from complex patient data is challenging because of the information heterogeneity and temporal dynamics. Long-Short Term Memory (LSTM) has been successfully used in many domains for processing sequential data, and recently applied for analyzing longitudinal patient records. The LSTM units are designed to handle data with constant elapsed times between consecutive elements of a sequence. Given that time lapse between successive elements in patient records can vary from days to months, the design of traditional LSTM may lead to suboptimal performance. In this paper, we propose a novel LSTM unit called Time-Aware LSTM (T-LSTM) to handle irregular time intervals in longitudinal patient records. We learn a subspace decomposition of the cell memory which enables time decay to discount the memory content according to the elapsed time. We propose a patient subtyping model that leverages the proposed T-LSTM in an auto-encoder to learn a powerful single representation for sequential records of patients, which are then used to cluster patients into clinical subtypes. Experiments on synthetic and real world datasets show that the proposed T-LSTM architecture captures the underlying structures in the sequences with time irregularities.