How a pandemic develops depends on many factors: the biology of the pathogen, its susceptibility to external influences such as humidity and temperature, the success of vaccination campaigns, but most of all it is determined by contacts between individuals: transmissions occur primarily when people have contact with each other.
The number of contacts thus gives us an insight into the course of the pandemic. Fewer contacts make it more difficult for the pathogen to spread through the population, whereas more contacts promote rapid spread.
In the Contacts Monitor, we show the number of contacts in Germany over time. In the following, we explain how these contacts are measured from GPS data of cell phones.
The data shown in the monitor is available here.
The analyses here are based on GPS trajectories of people living in Germany. However, what we receive from our cooperation partner is only the average number of contacts per day as well as their variance (in the section “What is a contact?” below, we explain how these observables relate to GPS trajectories). We do not have access to the GPS location data on which these measures are calculated.
The GPS location data is collected by the German company NETCHECK using cell phones. This type of data is commercially available. Such data is used, for example, by public transportation companies to measure traffic or by mobile network operators to improve cellular networks.
The GPS data is transmitted from the individual devices via a software development kit (SDK). This SDK is primarily used to evaluate the quality of mobile networks. It is installed on around 1.2 million devices in Germany, where all users have given their consent to collect such data for statistical purposes. Each device transmits up to several hundred locations per day. Of all these devices, about 600,000 are active every day.
Each measured GPS location is assigned to an area (“tile”) of 8 square meters. If two devices are on the same tile within 2 minutes, this event is defined as a contact between the two devices. If two devices meet more than once (for example, if a couple meets at home both in the morning and in the evening), the contact is still counted only once per day. Contacts are also not weighted by how much time the people spent near each other.
Each contact can be considered a connection between two devices in a network. NETCHECK aggregates these contacts to create a contact network for each day. In the network, all devices are connected to each other that had a contact on that day (where we assume that each device corresponds to one person).
The number of contacts between people influences the spread of an the disease. The measured contacts therefore give us an insight into the course of the pandemic.
The mean number of contacts indicates how many other people on average can potentially be infected by a single infected person.
The standard deviation of the number of contacts measures how much the contacts vary. A large variation means that there are individuals with a large number of contacts that could lead to “super-spreader” events.
It should be noted that the distribution of the number of contacts has a lower bound at zero. This means that a person can have many more contacts than the average number of contacts, but not much less (see simplified distribution above). Thus, a wide distribution means that there are especially many people with more contacts than average, not fewer contacts.
The contact network we measure here is of course only a sample. Not every person in Germany is represented in the data, and the measured network is therefore only a small section of the complete network. However, we use concepts from network theory to derive the mean and variance of the contacts of the complete unknown contact network from the sample. Details can be found in the following publication: Rüdiger et al..