Arbitrary Data as Images: Fusion of Patient Data Across Modalities and Irregular Intervals with Vision Transformers

TL;DR

This paper introduces ViTiMM, a vision transformer-based approach that visualizes diverse patient data modalities as images and text, simplifying multi-modal modeling and improving in-hospital mortality and phenotyping predictions.

Contribution

The paper presents a novel method that unifies multi-modal medical data as images and text for transformer-based modeling, reducing complexity and outperforming existing methods.

Findings

Outperforms state-of-the-art in-hospital mortality prediction

Effective across multiple modalities including images and signals

Simplifies multi-modal data integration for medical AI

Abstract

A patient undergoes multiple examinations in each hospital stay, where each provides different facets of the health status. These assessments include temporal data with varying sampling rates, discrete single-point measurements, therapeutic interventions such as medication administration, and images. While physicians are able to process and integrate diverse modalities intuitively, neural networks need specific modeling for each modality complicating the training procedure. We demonstrate that this complexity can be significantly reduced by visualizing all information as images along with unstructured text and subsequently training a conventional vision-text transformer. Our approach, Vision Transformer for irregular sampled Multi-modal Measurements (ViTiMM), not only simplifies data preprocessing and modeling but also outperforms current state-of-the-art methods in predicting in-hospital mortality and phenotyping, as evaluated on 6,175 patients from the MIMIC-IV dataset. The modalities include patient's clinical measurements, medications, X-ray images, and electrocardiography scans. We hope our work inspires advancements in multi-modal medical AI by reducing the training complexity to (visual) prompt engineering, thus lowering entry barriers and enabling no-code solutions for training. The source code will be made publicly available.