Fig 1.
Regional toll-free number daily incoming calls vs. daily infected time courses and wavelet analysis.
On the left, the smoothed (7-days amplitude moving average) and normalised time courses are displayed (toll-free number calls in blue, daily infected in red); on the right, WCS and MSWC chart is shown (see text for explanation).
Fig 2.
Regional NUE daily incoming calls vs. daily infected time courses and wavelet analysis.
On the left, the smoothed (7-days amplitude moving average) and normalised time courses are displayed (NUE calls in blue, daily infected in red); on the right, WCS and MSWC chart is shown (see text for explanation).
Fig 3.
Regional SOREU daily incoming calls vs. daily infected time courses and wavelet analysis.
On the left, the smoothed (7-days amplitude moving average) and normalised time courses are displayed (SOREU calls in blue, daily infected in red); on the right, WCS and MSWC chart is shown (see text for explanation).
Fig 4.
Daily number of tweets vs. regional daily infected time courses and wavelet analysis.
On the left, the smoothed (7-days amplitude moving average) and normalised time courses are displayed (tweets in blue, daily infected in red); on the right, WCS and MSWC chart is shown (see text for explanation).
Fig 5.
Daily number of likes vs. regional daily infected time courses and wavelet analysis.
On the left, the smoothed (7-days amplitude moving average) and normalised time courses are displayed (likes in blue, daily infected in red); on the right, WCS and MSWC chart is shown (see text for explanation).
Fig 6.
Daily number of retweets vs. regional daily infected time courses and wavelet analysis.
On the left, the smoothed (7-days amplitude moving average) and normalised time courses are displayed (retweets in blue, daily infected in red); on the right, WCS and MSWC chart is shown (see text for explanation).
Fig 7.
Daily number of replies vs. regional daily infected time courses and wavelet analysis.
On the left, the smoothed (7-days amplitude moving average) and normalised time courses are displayed (replies in blue, daily infected in red); on the right, WCS and MSWC chart is shown (see text for explanation).
Fig 8.
The trends of daily number of new tweets about emergency calls (blue line) and of replies (pink line) they sparked are shown here. The vertical green dotted lines indicate the principal episodes related to the lockdown policies in Italy. The orange box indicates the first peaks of activities, while the red circle highlights the highest peak in the number of replies.
Fig 9.
The trends of daily number of retweets (yellow line) and of likes (purple line) are shown here.
Fig 10.
Cross-correlation sequence estimate between NUE regional incoming calls and daily infected.
The blue lines represent the 90% confidence interval (CI) limits computed through a z-transformation. The maximum value of the cross-correlation function is depicted in red. Peak lag [CI]: -3 days [-8,1].
Fig 11.
Cross-correlation sequence estimate between SOREU regional incoming calls and daily infected.
The blue lines represent the 90% confidence interval (CI) limits computed through a z-transformation. The maximum value of the cross-correlation function is depicted in red. Peak lag [CI]: -5 days [-11,1].
Fig 12.
Cross-correlation sequence estimate between daily number of new tweets and daily infected.
The blue lines represent the 90% confidence interval (CI) limits computed through a z-transformation. The maximum value of the cross-correlation function is depicted in red. Peak lag [CI]: -6 days [-8,-2].
Table 1.
Sensitivity tests on the uncertainty in the location of the cross-correlation function peak.
Fig 13.
Cross-correlation analysis through a Monte Carlo simulation method for the regional NUE daily incoming calls vs. the daily infected time series.
A random phase test (see S1 File) has been performed (1,000 simulations) to compute the time lag to “align” the signals and the corresponding confidence interval (C.I.): time lag = -4 days (C.I. -11,1).
Fig 14.
Cross-correlation analysis through a Monte Carlo simulation method for the regional SOREU daily incoming calls vs. the daily infected time series.
A random phase test (see S1 File) has been performed (1,000 simulations) to compute the time lag to “align” the signals and the corresponding confidence interval (C.I.): time lag = -6 days (C.I. -13,1).
Fig 15.
Cross-correlation analysis through a Monte Carlo simulation method for the daily new tweets vs. the daily infected time series.
A random phase test (see S1 File) has been performed (1,000 simulations) to compute the time lag to “align” the signals and the corresponding confidence interval (C.I.): time lag = -3 days (C.I. -11,4).