Methods/Statistics
Differentially Private Modeling of Disease Transmission within Human Contact Networks Debanuj Nayak* Shlomi Hod Debanuj Nayak Iden Kalemaj Jason R. Gantenberg Thomas A. Trikalinos Adam Smith
Epidemiologic studies of infectious diseases often rely on models of contact networks to capture the complex interactions that govern disease spread. Data on how people interact may include sensitive information about sexual relationships or drug use behaviors. Protecting individual privacy while maintaining the scientific usefulness of the data is crucial.
Differential privacy (DP) is a mathematical framework that protects sensitive data by limiting the influence of a given person’s data on an aggregate statistic. In general, DP involves adding random noise—calibrated to control the tradeoff between privacy and accuracy—to the calculation. In the context of network data, where individuals are modeled as nodes of a network, node-DP safeguards both an individual’s data and their social connections.
We propose a privacy-preserving pipeline for simulating disease spread on network data which integrates node-DP with exponential random graph models (ERGMs). Our pipeline comprises three steps: (1) compute network statistics using node-DP, (2) fit an ERGM to these statistics to generate synthetic networks reflecting the structure of the original network, and (3) simulate disease spread on the synthetic networks using an agent-based model.
We evaluate the effectiveness of our approach using a simple Susceptible-Infected-Susceptible (SIS) disease model under multiple configurations. We compare simulated disease incidence and prevalence on networks generated with and without node-DP, based on egocentric sexual network data from the ARTNet study (a survey about HIV-related behaviors among MSM in the US). Our results show that the noise added for privacy is small relative to the other sources of error (sampling and modeling, for example). They suggest that, in principle, our pipeline can provide valuable epidemiologic insights while maintaining privacy. Future work will explore more complex networks and simulators.
