Table 1.
Characteristics of query topics screened for fit with coronavirus disease (COVID-19) case data.
| Search query category | Unique queriesa, n (%) | Correlation with national case rateb, mean | Associated lag time (days)c, mean | Activity weightingd |
| COVID-19 guidance | 32 (6.9) | 0.96 | 9.1 | 0.38 |
| COVID-19 news | 57 (12.3) | 0.96 | 8.3 | 1.00 |
| COVID-19 symptoms | 91 (19.7) | 0.94 | 8.9 | 0.41 |
| Medical treatments | 34 (7.3) | 0.93 | 10.1 | 0.31 |
| COVID-19 testing | 58 (12.5) | 0.89 | 5.4 | 0.11 |
| Medical care | 33 (7.1) | 0.89 | 7.2 | 0.60 |
| Nonspecific symptoms | 62 (13.4) | 0.89 | 6.8 | 0.57 |
| Economic effects | 28 (6.0) | 0.86 | 5.9 | 0.12 |
| Unrelated to illness | 51 (11.0) | 0.86 | 6.6 | 0.76 |
| Symptoms of other illnesses | 17 (3.7) | 0.84 | 8.3 | 0.77 |
aNumber of queries of each type in the query library (eg, the category “COVID-19 testing” would include the specific query “coronavirus test near me,” and the category “nonspecific symptoms” would include the query “cough”).
bExpressed as the inverse z-transformation of the averaged z-transformed correlations with in-sample national data.
cMean lag time between best-fitting query activity and confirmed case rate, in days.
dRelative mean search activity levels, normalized.