Spread of SARS-CoV-2 Coronavirus likely to be constrained by climate

Miguel B. Araujo; Babak Naimi

We used the area under curve (AUC) of receiver operating characteristic (ROC) plot and the true skill statistic (TSS) to measure the predictive performance of models (54) . A ROC curve plots sensitivity values (true positive fraction) on the yaxis against '1 -specificity' values (false positive fraction) for all thresholds on the x-axis. AUC is a threshold-independent metric that varies from 0 to 1 and provides a single measure of model performance. AUC values under 0.5 indicate discrimination worse than expected by chance; a score of 0.5 implies random predictive discrimination; and a score of 1 indicates perfect discrimination. TSS is calculated as "sensitivity + specificity -1" and ranges from -1 to +1, where +1 indicates perfect agreement, a value of 0 implies agreement expected by chance, and a value of less than 0 indicates agreement lower than expected by chance. We then used the ensemble of 200 models to calculate and project a consensus distribution of climate suitability for the spread of SARS-CoV-2, for each month, across the globe. Consensus was achieved through AUC-weighted mean across all models (21) . The assessments of model performance on cross-validated samples are reported in Supplementary Figure S3 .