Interpretable Neural Temporal Point Processes for Modelling Electronic Health Records

TL;DR

This paper introduces inf2vec, an interpretable neural temporal point process model for electronic health records that explicitly captures event dependencies, improving event prediction and interpretability.

Contribution

The paper proposes a novel interpretable framework for neural temporal point processes, directly parameterizing event influences for better understanding and end-to-end learning.

Findings

Outperforms existing models in event prediction accuracy.

Effectively learns type-type influence relationships.

Provides interpretable insights into event dependencies.

Abstract

Electronic Health Records (EHR) can be represented as temporal sequences that record the events (medical visits) from patients. Neural temporal point process (NTPP) has achieved great success in modeling event sequences that occur in continuous time space. However, due to the black-box nature of neural networks, existing NTPP models fall short in explaining the dependencies between different event types. In this paper, inspired by word2vec and Hawkes process, we propose an interpretable framework inf2vec for event sequence modelling, where the event influences are directly parameterized and can be learned end-to-end. In the experiment, we demonstrate the superiority of our model on event prediction as well as type-type influences learning.