Step by Step
The distribution of daily step counts
In addition to the daily average resting heart rate, we can obtain the daily step count from connected wearable fitness devices. Together with the resting heart rate, the step count is an important parameter for this study. As you can read in our post, The Pulse of the Nation, deviations from the average resting heart rate could indicate fever. However, resting heart rate may also deviate for various unrelated reasons.
By taking the step count into consideration as well as the resting heart rate, we can obtain a stronger indicator for fever: If your daily step count is much lower than your average step count, while your resting heart rate is significantly higher than baseline, it is more likely that fever is the underlying culprit. Thus, we obtain stronger results when combining step count and heart rate in our analysis.
However, before combining both parameters into one analysis, the step count itself needs to be analyzed on its own. By doing this, important variations due to weekday patterns or weather factors can be identified first. Moreover, step counts may even vary regionally between rural and urban areas. Taking all this into consideration, we aim to establish a solid reference baseline for further analyses.
Step count statistics for Germany and the federal states
To determine the typical daily step count and variation of donors in Germany, we plotted the the distribution of average daily step counts for all of Germany and each of the federal states:
As can be seen in the figure above, typical daily values fall between 5,000 and 7,000 steps, with a fairly broad variation. Some donors even show a significantly higher step count. Variations between different German states, on the other side, are rather insignificant, leading to a uniformly shaped step count curve.
The step count map
Similar to the Resting Heart Rate Map, we can determine the average step count per district in Germany. The result is shown in the following map.
The map depicts the average daily step count, computed for each district during the period from April 12 to May 18, 2020. The variation around the mean extends between 6,500 and 8,500 steps a day. In some regions, e.g. in Thuringia’s Kyffhäuserkreis and Schmalkalden-Meinigen, the step count is significantly elevated compared to the nation-wide average daily step count. Also, there is a clear difference between urban and rural areas in general. Donors in urban areas tend to take less steps a day than donors in rural areas.
Temporal stability of the step count parameter
Now, let’s look at the temporal stability of the step count. The below figure depicts the average step count for all of Germany as a function of time between April 12 and May 18, 2020.
We can see that the daily step count follows roughly a constant curve. However, long-term tends cannot be predicted. The daily average is between 6,000 and 7,000 steps. Compared to the average daily resting heart rate, the step count displays higher variation (see Figure 2 in this post)). Moreover, weekends, especially Saturdays, are more prone to higher step counts than other weekdays.
Average daily step count in German states
Now the question arises whether the number of steps of the donors differs regionally. The figure below shows the average daily step count of donors for each of the German states.
In this figure, we can see that donors walk or run more frequently in some states as compared to others. In Hamburg and Bremen, for instance, the average step count is typically lower, whereas donors in Thuringia and Saxony are more “on-the-go” than donors in other states.
In addition, this figure clearly shows that the average number of steps on weekends, specifically on Saturdays, is higher than on weekdays.
Variability of step counts
The average number of steps per day, whether nation-wide or district-wide, is an important measure for this study. However, it is not the only statistically relevant value. We are also interested in the variability in the number of steps, which is the magnitude of the distribution. The variation can be determined via another statistical parameter, the standard deviation, which determines how far individual values are from the mean.
The following figure depicts the standard deviation of the step count as a function of time for each federal state.
This figure shows that, the variability in the number of steps increases significantly on weekends and public holidays compared to weekdays. This is seen across all federal states.
Another important statistical parameter that requires attention is the coefficient of variation, which describes the relation between standard deviation and statistical mean. It is sometimes referred to as the “relative standard deviation”. The figure below depicts the coefficient of variation as a function of time for Germany.
The following analyses include combining the resting heart rate and the step count in order to identify statistical and dynamic effects. In the end, both parameters will comprise the basis of a fever-detection algorithm, which will be applied at the district level in order to detect sudden spikes of fever-like symptoms.