[Paper Review] COVI-AgentSim: an Agent-based Model for Evaluating Methods of Digital Contact Tracing

Prateek K. Gupta, Tegan Maharaj+27 more|arXiv (Cornell University)|Oct 29, 2020

COVID-19 Digital Contact Tracing4 citations

TL;DR

COVI-AgentSim is an agent-based simulation model that evaluates digital contact tracing (DCT) methods for SARS-CoV-2 by integrating virology, mobility, and social contact data. It demonstrates that feature-based contact tracing (FCT) significantly reduces disability-adjusted life years (DALYs) and improves cost-effectiveness compared to standard binary contact tracing (BCT), even at lower adoption rates.

ABSTRACT

The rapid global spread of COVID-19 has led to an unprecedented demand for effective methods to mitigate the spread of the disease, and various digital contact tracing (DCT) methods have emerged as a component of the solution. In order to make informed public health choices, there is a need for tools which allow evaluation and comparison of DCT methods. We introduce an agent-based compartmental simulator we call COVI-AgentSim, integrating detailed consideration of virology, disease progression, social contact networks, and mobility patterns, based on parameters derived from empirical research. We verify by comparing to real data that COVI-AgentSim is able to reproduce realistic COVID-19 spread dynamics, and perform a sensitivity analysis to verify that the relative performance of contact tracing methods are consistent across a range of settings. We use COVI-AgentSim to perform cost-benefit analyses comparing no DCT to: 1) standard binary contact tracing (BCT) that assigns binary recommendations based on binary test results; and 2) a rule-based method for feature-based contact tracing (FCT) that assigns a graded level of recommendation based on diverse individual features. We find all DCT methods consistently reduce the spread of the disease, and that the advantage of FCT over BCT is maintained over a wide range of adoption rates. Feature-based methods of contact tracing avert more disability-adjusted life years (DALYs) per socioeconomic cost (measured by productive hours lost). Our results suggest any DCT method can help save lives, support re-opening of economies, and prevent second-wave outbreaks, and that FCT methods are a promising direction for enriching BCT using self-reported symptoms, yielding earlier warning signals and a significantly reduced spread of the virus per socioeconomic cost.

Motivation & Objective

• To develop a high-fidelity agent-based simulator for evaluating digital contact tracing (DCT) methods in realistic pandemic scenarios.
• To address the limitations of binary contact tracing (BCT), which relies solely on test results and suffers from false negatives and delayed detection.
• To evaluate whether feature-based contact tracing (FCT), which uses self-reported symptoms and other individual features, can improve early warning and transmission control.
• To perform cost-benefit analyses comparing no DCT, BCT, and FCT in terms of lives saved, productivity loss, and DALYs averted.
• To provide a testbed for optimizing real-world DCT implementations with evidence-based public health decisions.

Proposed method

• COVI-AgentSim uses an agent-based compartmental model simulating individual-level interactions, disease progression, and mobility patterns based on empirical data.
• The model incorporates virological parameters such as viral load dynamics, incubation periods, and infectiousness profiles derived from real-world studies.
• It simulates social contact networks with realistic age- and location-based mixing patterns to reflect real-world transmission dynamics.
• The simulator evaluates two DCT strategies: (1) binary contact tracing (BCT), which issues binary quarantine recommendations based on test results; and (2) heuristic feature-based contact tracing (FCT), which assigns graded quarantine recommendations using symptom reports and other features.
• Cost-benefit analysis is performed using disability-adjusted life years (DALYs) and total productive hours lost (TPL), with adjustments for remote work capacity.
• Model validation is performed by comparing simulated outcomes to real-world data from the GBD 2017 study and sensitivity analysis confirms robustness across parameter variations.

Experimental results

Research questions

• RQ1 How do different DCT strategies—binary contact tracing (BCT) and feature-based contact tracing (FCT)—compare in reducing SARS-CoV-2 transmission under varying adoption rates?
• RQ2 To what extent does FCT reduce disability-adjusted life years (DALYs) compared to BCT and no tracing, especially in terms of early detection and intervention?
• RQ3 What is the socioeconomic cost-effectiveness of FCT versus BCT, measured in productive hours lost per DALY averted?
• RQ4 How robust are the relative performance differences between BCT and FCT across diverse epidemiological and behavioral parameter settings?
• RQ5 Can the simulator reproduce realistic pandemic dynamics and serve as a reliable benchmark for evaluating future DCT interventions?

Key findings

• All DCT methods significantly reduce disease spread compared to no contact tracing, with FCT showing a consistent advantage over BCT across a range of adoption rates.
• Feature-based contact tracing (FCT) averts 44.90% more DALYs than no tracing, reducing total DALYs from 129.42 to 71.31, while increasing total productive hours lost (TPL) to $2.122M.
• FCT reduces DALYs by 44.90% at the cost of a 54.89% increase in TPL, indicating a favorable trade-off in health outcomes per unit of economic cost.
• The relative performance of FCT over BCT remains robust across sensitivity analyses, suggesting consistent advantages regardless of parameter variation.
• FCT outperforms BCT in cost-effectiveness, delivering significantly more DALYs averted per unit of socioeconomic cost (productive hours lost).
• The model successfully reproduces real-world pandemic dynamics when validated against GBD 2017 data, confirming its fidelity and utility as a benchmarking tool.