Automated Fusion of Multimodal Electronic Health Records for Better Medical Predictions

TL;DR

This paper introduces AutoFM, a neural architecture search framework that automatically designs optimal models for multi-modal EHR data, significantly improving predictive performance over existing methods.

Contribution

The paper presents a novel NAS framework, AutoFM, specifically tailored for multi-modal EHR data, automating model design and enhancing predictive accuracy.

Findings

AutoFM outperforms state-of-the-art methods in EHR prediction tasks.

AutoFM effectively discovers meaningful and diverse network architectures.

The framework demonstrates significant performance improvements on real-world data.

Abstract

The widespread adoption of Electronic Health Record (EHR) systems in healthcare institutes has generated vast amounts of medical data, offering significant opportunities for improving healthcare services through deep learning techniques. However, the complex and diverse modalities and feature structures in real-world EHR data pose great challenges for deep learning model design. To address the multi-modality challenge in EHR data, current approaches primarily rely on hand-crafted model architectures based on intuition and empirical experiences, leading to sub-optimal model architectures and limited performance. Therefore, to automate the process of model design for mining EHR data, we propose a novel neural architecture search (NAS) framework named AutoFM, which can automatically search for the optimal model architectures for encoding diverse input modalities and fusion strategies. We conduct thorough experiments on real-world multi-modal EHR data and prediction tasks, and the results demonstrate that our framework not only achieves significant performance improvement over existing state-of-the-art methods but also discovers meaningful network architectures effectively.