Table 5.

Correlations between valid tweets and emergency department ILI^a rates by city.^b

	1. All tweets	2. Number of all tweets	3. All tweets	4. Valid tweets	5. Number of valid tweets	6. Valid tweets	7. Fisher’s z transformation
	r		P	r	r	P	P
Boston	.23	17,370	.411	.61	3813	.016	<.001
Chicago	.51	21,655	.017	.80	5116	<.001	<.001
Cleveland	.68	7152	<.001	.75	1497	<.001	<.001
Columbus	.62	3288	.002	.87	1034	<.001	<.001
San Diego	.80	8002	<.001	.88	1808	<.001	<.001
Seattle	.72	9735	<.001	.82	2941	<.001	<.001

^aILI: influenza-like illness

^bCorrelation coefficients between all tweets and valid tweets, as identified by the machine-learning classifier, with emergency department ILI rates for each city. Comparisons between tweets and ILI began in Weeks 40-41 (weeks starting September 29, 2013 to starting October 6, 2013) as ILI data became available by city and ended in Week 9 (ending March 1, 2014).