Contact patterns among high school students

Face-to-face contacts between individuals contribute to shape social networks and play an important role in determining how infectious diseases can spread within a population. It is thus important to obtain accurate and reliable descriptions of human contact patterns occurring in various day-to-day life contexts. Recent technological advances and the development of wearable sensors able to sense proximity patterns have made it possible to gather data giving access to time-varying contact networks of individuals in specific environments. Here we present and analyze two such data sets describing with high temporal resolution the contact patterns of students in a high school. We define contact matrices describing the contact patterns between students of different classes and show the importance of the class structure. We take advantage of the fact that the two data sets were collected in the same setting during several days in two successive years to perform a longitudinal analysis on two very different timescales. We show the high stability of the contact patterns across days and across years: the statistical distributions of numbers and durations of contacts are the same in different periods, and we observe a very high similarity of the contact matrices measured in different days or different years. The rate of change of the contacts of each individual from one day to the next is also similar in different years. We discuss the interest of the present analysis and data sets for various fields, including in social sciences in order to better understand and model human behavior and interactions in different contexts, and in epidemiology in order to inform models describing the spread of infectious diseases and design targeted containment strategies.

💡 Research Summary

This paper presents a detailed investigation of face‑to‑face contact patterns among high‑school students using two high‑resolution data sets collected in the same French high school (Lycée Thiers, Marseille) during 2011 and 2012. Wearable badges from the SocioPatterns platform recorded proximity events every 20 seconds, defining a contact when two devices exchanged radio packets within a 1–1.5 m range. In 2011, three “classes préparatoires” (118 students) were monitored over four school days; in 2012, five such classes (180 students) were monitored over seven days. The total number of contact events recorded in 2012 was 19,774, with a cumulative duration of about 250 hours.

The authors first characterize the basic statistics of contacts. The distribution of contact durations is heavy‑tailed: the mean duration is 44 seconds, but the squared coefficient of variation (CV²) equals 5, indicating extreme heterogeneity. About 88 % of contacts last less than one minute, yet more than 1 % exceed five minutes, suggesting that a small fraction of long contacts could dominate disease transmission. Inter‑contact times are also broadly distributed, close to a power‑law with exponent < 2.

To capture structural patterns, the authors aggregate the temporal network into weighted graphs over various time windows (daily, whole‑period). Nodes represent individuals; edges are weighted by the total contact time (wᵢⱼ) and also by the number of distinct contact events (nᵢⱼ). From these graphs they compute degree (kᵢ) and strength (sᵢ=∑ⱼ wᵢⱼ) for each student, and analyse their distributions. Both degree and strength show high variability (CV² > 1), indicating the presence of potential “super‑spreader” individuals.

A central focus is the role of the class organization. The authors construct contact matrices at the class level: Eₓᵧ (number of edges between classes X and Y), ρₓᵧ (edge density relative to the maximum possible), Nₓᵧ (total number of contacts), and Wₓᵧ (total contact time). The matrices display a pronounced block structure: intra‑class densities ρₓₓ are typically 0.6–0.8, while inter‑class densities ρₓᵧ are below 0.1. This confirms that the class division is the dominant factor shaping the contact network, with very limited mixing between different classes.

Temporal stability is assessed using two similarity measures. At the individual level, the cosine similarity σ₁,₂(i) compares the weighted neighbor vectors of student i on two different days; values range from 0 (completely different partners) to 1 (identical partners with proportional contact times). Daily similarities are high (average σ≈0.78–0.84), indicating that most students retain the same set of contacts from one day to the next. At the matrix level, the similarity σ_{A,B} compares whole contact matrices from different days or years; values exceed 0.85 both for intra‑year day‑to‑day comparisons and for the inter‑year comparison (2011 vs. 2012). Thus, despite a complete turnover of individual students between the two years, the overall pattern of contacts remains remarkably stable.

Gender differences are examined as well. Although the proportion of males and females varies across classes, no statistically significant differences are found in either contact frequency or total contact time, suggesting that gender does not strongly influence contact behavior in this age group.

The authors discuss the implications for epidemiology and social science. The strong class‑based segregation implies that interventions targeting whole classes (e.g., class‑level quarantine, targeted vaccination) could be highly effective in limiting outbreak spread. However, the heavy‑tailed contact‑duration distribution and the presence of highly connected individuals mean that models assuming homogeneous mixing would underestimate the speed and reach of transmission. From a sociological perspective, the high intra‑class contact density reflects the tight social cohesion typical of high‑school cohorts, while the low inter‑class mixing underscores the limited cross‑group interactions.

Finally, the data set, together with analysis scripts, is made publicly available on the SocioPatterns website, providing a valuable resource for researchers studying temporal networks, disease dynamics, or human behavior in educational settings.

In summary, this work demonstrates that high‑resolution wearable sensors can capture detailed, reproducible contact patterns in a high‑school environment. The findings highlight the dominant role of class structure, the stability of contact patterns across days and years, and the importance of accounting for contact heterogeneity in models of disease spread.