Out-of-Distribution Generalized Dynamic Graph Neural Network for Human Albumin Prediction

TL;DR

This paper proposes a framework named Out-of-Distribution Generalized Dynamic Graph Neural Network for Human Albumin Prediction (DyG-HAP), which is able to provide accurate albumin predictions for Intensity Care Unit (ICU) patients during hospitalization and proposes a disentangled dynamic graph attention mechanism to capture and disentangle patterns.

Abstract

Human albumin is essential for indicating the body's overall health. Accurately predicting plasma albumin levels and determining appropriate doses are urgent clinical challenges, particularly in critically ill patients, to maintain optimal blood levels. However, human albumin prediction is non-trivial that has to leverage the dynamics of biochemical markers as well as the experience of treating patients. Moreover, the problem of distribution shift is often encountered in real clinical data, which may lead to a decline in the model prediction performance and reduce the reliability of the model's application. In this paper, we propose a framework named Out-of-Distribution Generalized Dynamic Graph Neural Network for Human Albumin Prediction (DyG-HAP), which is able to provide accurate albumin predictions for Intensity Care Unit (ICU) patients during hospitalization. We first model human albumin prediction as a dynamic graph regression problem to model the dynamics and patient relationship. Then, we propose a disentangled dynamic graph attention mechanism to capture and disentangle the patterns whose relationship to labels under distribution shifts is invariant and variant respectively. Last, we propose an invariant dynamic graph regression method to encourage the model to rely on invariant patterns to make predictions. Moreover, we propose a dataset named Albumin level testing and nutritional dosing data for Intensive Care (ANIC) for evaluation. Extensive experiments demonstrate the superiority of our method compared to several baseline methods in human albumin prediction.

Figures (5)

• Figure 1: Changes in the distribution of biochemical markers over time.
• Figure 2: The framework of our proposed Out-of-Distribution Generalized Dynamic Graph Neural Network for Human Albumin Prediction (DyG-HAP). (Left) We collect the ANIC dataset that has various patient features, including demographic characteristics, nutritional support, biochemical markers, etc. (Middle) We model the human albumin problem as dynamic graph regression and construct a dynamic graph with varying features and structures to simultaneously consider the dynamic patient relationship and attributes. (Right) We propose a model composed of a disentangled dynamic graph attention that captures and disentangles the invariant and variant patterns, and an invariant dynamic graph regression method to encourage the model to rely on invariant patterns to make predictions.
• Figure 3: Comparisons of different methods in terms of mean absolute error (MAE) at each time.
• Figure 4: A showcase of the albumin predictions of different methods.
• Figure 5: The importance of patient features given by our model.