Differentiable Agent-based Epidemiology

07/20/2022
by   Ayush Chopra, et al.
7

Mechanistic simulators are an indispensable tool for epidemiology to explore the behavior of complex, dynamic infections under varying conditions and navigate uncertain environments. ODE-based models are the dominant paradigm that enable fast simulations and are tractable to gradient-based optimization, but make simplifying assumptions about population homogeneity. Agent-based models (ABMs) are an increasingly popular alternative paradigm that can represent the heterogeneity of contact interactions with granular detail and agency of individual behavior. However, conventional ABM frameworks are not differentiable and present challenges in scalability; due to which it is non-trivial to connect them to auxiliary data sources easily. In this paper we introduce GradABM which is a new scalable, fast and differentiable design for ABMs. GradABM runs simulations in few seconds on commodity hardware and enables fast forward and differentiable inverse simulations. This makes it amenable to be merged with deep neural networks and seamlessly integrate heterogeneous data sources to help with calibration, forecasting and policy evaluation. We demonstrate the efficacy of GradABM via extensive experiments with real COVID-19 and influenza datasets. We are optimistic this work will bring ABM and AI communities closer together.

READ FULL TEXT

page 2

page 4

research
10/09/2021

DeepABM: Scalable, efficient and differentiable agent-based simulations via graph neural networks

We introduce DeepABM, a framework for agent-based modeling that leverage...
research
07/03/2023

Some challenges of calibrating differentiable agent-based models

Agent-based models (ABMs) are a promising approach to modelling and reas...
research
03/23/2021

Differentiable Agent-Based Simulation for Gradient-Guided Simulation-Based Optimization

Simulation-based optimization using agent-based models is typically carr...
research
05/24/2023

Bayesian calibration of differentiable agent-based models

Agent-based modelling (ABMing) is a powerful and intuitive approach to m...
research
05/10/2022

On learning agent-based models from data

Agent-Based Models (ABMs) are used in several fields to study the evolut...
research
06/28/2022

Rethinking Optimization with Differentiable Simulation from a Global Perspective

Differentiable simulation is a promising toolkit for fast gradient-based...

Please sign up or login with your details

Forgot password? Click here to reset