Spatial-Temporal Networks for Antibiogram Pattern Prediction

Abstract

An antibiogram is a periodic summary of antibiotic resistance results of organisms from infected patients to selected antimicrobial drugs. Antibiograms help clinicians to understand regional resistance rates and select appropriate antibiotics in prescriptions. In practice, significant combinations of antibiotic resistance may appear in different antibiograms, forming antibiogram patterns. Such patterns may imply the prevalence of some infectious diseases in certain regions. Thus it is of crucial importance to monitor antibiotic resistance trends and track the spread of multi-drug resistant organisms. In this paper, we propose a novel problem of antibiogram pattern prediction that aims to predict which patterns will appear in the future. Despite its importance, tackling this problem encounters a series of challenges and has not yet been explored in the literature. First of all, antibiogram patterns are not i.i.d as they may have strong relations with each other due to genomic similarities of the underlying organisms. Second, antibiogram patterns are often temporally dependent on the ones that are previously detected. Furthermore, the spread of antibiotic resistance can be significantly influenced by nearby or similar regions. To address the above challenges, we propose a novel Spatial-Temporal Antibiogram Pattern Prediction framework, STAPP, that can effectively leverage the pattern correlations and exploit the temporal and spatial information. We conduct extensive experiments on a real-world dataset with antibiogram reports of patients from 1999 to 2012 for 203 cities in the United States. The experimental results show the superiority of STAPP against several competitive baselines.

Keywords: antibiogram patterns; antibiotic resistance; attention mechanism; spatial-temporal learning.

Fig. 1:

An example of antibiogram pattern prediction. This example includes antibiogram patterns appearing in four timesteps from $t=1$ to $t=4$ for three regions. The blue square, the green star, the orange triangle, and the red circle represent four distinct antibiogram patterns. The goal of the three regions is to predict which antibiogram patterns will appear in timestep $t=5$.

Fig. 2:

An overview of STAPP. STAPP first constructs antibiogram pattern graphs ${\{{𝒢}_t^{\left(k\right)}\}}_{t=1}^T$ from $t=1$ to $t=T$ for each region $r^{\left(k\right)}$. It then employs an antibiogram pattern graph convolution module, a temporal attention module, and a spatial graph convolution module to predict antibiogram pattern presences in $t=T+1$ using the information from ${\{{𝒢}_t^{\left(k\right)}\}}_{t=1}^T$ as well as $r^{\left(k\right)}$,s neighboring regions.

Fig. 3:

F1 score with different values of $\delta$ and $\kappa$.

Fig. 4:

Comparison of ${\mathbf{A}}_{kq}^{ℛ}$ in STAPP-D and STAPP-J.