Abstract
Traditional vaccination strategies often target individuals based on their total number of contacts (Degree-based Vaccination) or at random. While effective in theory, contact-based strategies are difficult to implement due to privacy concerns and data inaccuracies. This paper introduces Individual’s Movement-based Vaccination (IMV), a strategy that prioritizes individuals based on the types of locations they frequent. Using a city-wide mobility dataset, simulations demonstrate that IMV achieves efficiency comparable to ideal contact-based methods while requiring significantly less data and offering better privacy protections.
Introduction
Controlling a pandemic requires identifying and vaccinating "super-spreaders" early. Network theory suggests that Degree-based Vaccination (DV), which targets individuals with the highest number of contacts, is the most effective approach. However, DV is impractical in the real world; it requires precise, invasive tracking of every person an individual meets. Conversely, Random Vaccination (RV) is easy to implement but inefficient, requiring high coverage to be effective. This study proposes a middle ground: Individual’s Movement-based Vaccination (IMV), which focuses on "where" people go rather than "who" they meet.
Methods
The IMV strategy categorizes locations into six classes based on their potential for disease transmission, determined by estimated contact sizes and typical venue types.
Individuals are assigned a ranking score based on the frequency of their visits to these location classes. To test this, researchers used a city-wide mobility dataset from Momo, a location-based social app, to construct a "Dense Same Place Different Time" (DDT) contact network involving over 360,000 users. This network accounts for both direct and indirect transmission (e.g., lingering aerosols). Simulations compared IMV against RV, DV, and Acquaintance-based Vaccination (AV) across various vaccination rates and R-values.
Results
The simulations revealed that IMV is nearly as efficient as the "gold standard" DV strategy and vastly outperforms RV and AV. At a 2% vaccination rate, IMV reduced outbreak sizes by 99.0%, compared to 99.6% for DV and only 19.0% for RV.
A critical metric was the vaccination coverage required to keep an outbreak under 100 infections. IMV required vaccinating only 1.6% of the population to reach this threshold, whereas RV required 86.0% and AV required 65.0%.
Furthermore, IMV proved resilient to more infectious variants. When the R-value was increased to 1.7, IMV still reduced outbreak sizes by a factor of eight compared to random vaccination.
Discussion
IMV offers several practical advantages. It is cognitively easier for individuals to recall the types of places they visited than the specific people they encountered, making data collection via surveys more reliable. By focusing on location categories rather than specific names or contacts, it also provides a more privacy-preserving framework. Most importantly, it captures indirect transmission risks at "hotspots" that traditional contact tracing might miss.
While the study used synthetic density adjustments to compensate for sparse app data, the results suggest that IMV could be a powerful tool for public health officials, particularly in the early stages of a pandemic when vaccine supplies are limited. Future work could explore the strategy's effectiveness against other respiratory illnesses.
Summary updated