Contact tracing faltered sooner in Seoul than Busan when cases went untraced

A South Korean city simulation found that missing infected people mattered more than missing individual contact events, with estimated limits differing between Seoul and Busan.

Editorial illustration — not from the study.

A computer simulation of cities in South Korea found that missing information can affect contact tracing in different ways. Leaving infected people out of the tracing process had a sharper effect on epidemic spread than missing individual contact events.

The simulated threshold for contact-tracing effectiveness was about 4% of infected people omitted in Seoul, compared with about 10% in Busan. The results suggest that the amount of information a tracing system needs may depend on a city’s population and mobility structure.

How missing data changed spread

The researchers modeled two kinds of missing information. In one, infected people were omitted from the tracing process; in the other, some contact events were missed even when the infected people were identified.

Omitting infected people caused substantially stronger and more abrupt changes in the transmission network and epidemic outcomes. Missing contact events produced more gradual effects. In the simulations, contact-tracing effectiveness broke down beyond an omission rate of approximately 4% in Seoul and about 10% in Busan.

As either type of missing information increased, the transmission network’s diameter also grew. A smaller network diameter was associated with contact tracing that more effectively limited the depth of transmission chains.

Why city structure matters

The findings suggest that contact-tracing systems should not rely on one universal completeness threshold. A level of missing information that can be tolerated in one city may be associated with a sharper loss of effectiveness in another, depending on population and mobility structure. The results also indicate that identifying infected people may be especially important, because omissions at that stage had a stronger effect than missing individual contact events in the simulations.

Model evidence and limits

The study used a stochastic agent-based computer model, populated with sociodemographic information from metropolitan areas in South Korea, to simulate epidemic spread under different contact-tracing data losses. It was not a clinical study or a direct evaluation of contact-tracing programs. The threshold estimates come from the model scenarios described in the abstract, so their applicability to other cities, diseases or tracing systems is uncertain.

Reliable AI partnerships depend on people, not just stronger models

A survey of human-AI collaboration argues that more capable AI systems do not automatically produce reliable partnerships. It identifies human-centered design, governance and attention to high-stakes uses as central to making these systems trustworthy and beneficial.

9 hours ago·6 views

AI & Computing

RNA splicing tools can follow irrelevant clues and miss RNA’s shape

Deep-learning tools for predicting RNA splicing can rely on genomic clues unrelated to splicing and miss important effects of RNA structure, a study finds. These weaknesses were linked to systematic errors, especially when the tools evaluated sequences that differed from those used for training.

20 hours ago·5 views

AI & Computing

Better-trained AI finds more flaws in unfamiliar software projects

AI vulnerability detectors often perform less well on software projects outside their training data. A study finds that cleaner, more varied training data and an encoder-based design improved detection, including a reported 6.8% gain in recall on the BigVul benchmark.

2 days ago·34 views