[论文解读] The missing links: Evaluating contact tracing with incomplete data in large metropolitan areas during an epidemic

Min-Kyung Chae, Woo-Sik Son+1 more|arXiv (Cornell University)|Jan 21, 2026

COVID-19 epidemiological studies被引用 0

一句话总结

该论文使用对首尔和釜山的高分辨率基于代理的建模来量化信息丢失对手工接触者追踪效果的影响，识别城市特有的阈值与传播结构变化。

ABSTRACT

Contact tracing (CT) plays a pivotal role in controlling early epidemic spread, particularly when a novel infectious disease emerges. However, the quantitative impact of missing information -- such as untraced cases or unnotified contacts -- on the effectiveness of CT remains insufficiently understood. Using a stochastic agent-based model with sociodemographics from metropolitan areas in South Korea, we simulate how different forms of information loss affect epidemic spreading dynamics. We construct information-loss scenarios based on two types: infector-omission (IO) and contact-omission (CO), including selective (SCO) and uniform (UCO) scenarios; IO corresponds to the omission of infected individuals (nodes) from the tracing process, leading to the loss of all movement trajectories and downstream transmission links originating from them, whereas CO corresponds to the omission of specific contact events (edges), in which infected individuals are identified but some of their transmission links fail to be detected or notified. The sensitivity of epidemic dynamics to increasing omission rates differs markedly between the two types: IO scenarios exhibit substantially stronger and more abrupt changes in transmission structure and epidemic outcomes, whereas CO scenarios produce more gradual effects. In both scenarios, the magnitude of these effects varies across cities, with a lower-population city (Busan) showing greater tolerance to information loss than the largest city (Seoul), underscoring the importance of regional tailoring in CT strategies. Both IO and CO scenarios also lead to an increase in the transmission network diameter as information loss grows, indicating that a small network diameter reflects effective contact tracing that limits the depth of transmission chains.

研究动机与目标

• 理解手工 CT 中缺失信息如何影响疫情控制的动机。
• 开发具有多层社会网络的高分辨率 ABM，以模拟首尔和釜山并评估信息丢失下的 CT。
• 比较两种信息丢失类型（传染者遗漏 IO 与接触遗漏 CO）并推导对政策有用的阈值。
• 为针对区域人口统计与流动模式定制 CT 策略提供城市特异性指南。

提出的方法

• 从人口普查数据构建合成的首尔和釜山人口并投射到完整城市规模。
• 构建包含家庭、工作场所、学校、友谊和本地社区的多层接触网络。
• 使用扩展的 SEIR 框架建模疾病进展，纳入无症状病例与随机传播。
• 将手工 CT 实现为轨迹重建和递归追踪，包含 IO、SCO 和 UCO 场景，以及检测/隔离动态。
• 以缺失率作为函数，量化疫情结果与传播网络拓扑（直径）等指标。
• 每种情景运行 100 次随机仿真以评估鲁棒性。

实验结果

研究问题

• RQ1 在手工 CT 下，传染者遗漏 IO 如何影响疫情规模、时序与传播网络结构？
• RQ2 不同的接触遗漏 CO 场景（SCO 与 UCO）如何影响 CT 有效性与暴发动态？
• RQ3 是否存在城市特定的 CT 有效性阈值，且首尔与釜山在这些阈值上有何差异？
• RQ4 伴随信息丢失增加，会带来哪些网络拓扑变化（如传播网络直径）？
• RQ5 为设计对不完整数据具有鲁棒性的 CT 系统提供哪些政策指导？

主要发现

• IO 会导致控制效果的显著转变，虚拟首尔的阈值约为 4%，虚拟釜山约为 10%，超过该阈值后感染迅速上升。
• CO 场景带来更为渐进的疫情峰值延迟和相对较小的传播规模增加，与 IO 相比变化较小。
• 在城市间的共同点是，信息不全的轨迹重构（IO）会加深传播链，随着遗漏增大，定向传播网络直径增大。
• 选择性 CO（SCO）和均匀 CO（UCO）都会降低 CT 绩效，但 IO 仍是推动疫情扩展的主导因素。
• 在低遗漏率下 CT 仍然有效；即使轨迹追踪损失 modest 也可能将动态推向持续性疫情增长。
• 结果基于合成群体，并承认相对于真实 CT 日志的验证及固定延迟假设的局限性。