Fig. 2: Illustrative tables relating symptoms status with COVID-19 status. | Nature Machine Intelligence

Fig. 2: Illustrative tables relating symptoms status with COVID-19 status.

From: Audio-based AI classifiers show no evidence of improved COVID-19 screening over simple symptoms checkers

Fig. 2: Illustrative tables relating symptoms status with COVID-19 status.

a, Symptoms-based enrolment, where individuals who are COVID+ are preferentially recruited on the basis of symptoms (percentages are calculated from the entire sample of individuals recruited into this study). b, General population enrolment on the basis of random sampling from an illustrative general population with a COVID-19 prevalence of 2%, where symptomatic individuals make up 20% and 65% of COVID– and COVID+ subpopulations, respectively. c, Matched enrolment, where the number of individuals who are COVID– and COVID+ is the same for each particular symptoms profile within the symptomatic and asymptomatic subgroups (percentages shown are for the matched test set in the current study). For each type of enrolment, the diagnostic accuracies of the resulting symptoms-only COVID-19 classifier are shown below the table: ρ, mutual information (MI), sensitivity, specificity and AUC.

Back to article page