Predicting Whom to Test is More Important Than More Tests - Modeling the Impact of Testing on the Spread of COVID-19 Virus By True Positive Rate Estimation

Paul F Rodriguez

The contour plot in Figure 3 shows maximum values reached during the simulation, and a sample growth curve with a peak value of 188K infected persons, and a value on day 58 near the currently reported number (~50K). The true number is likely higher, but it gives a starting point to consider tradeoffs in parameters. The parameters for the sample growth curve were beta =0.42 (same as Figure 2 ) and TP-infpt 0.15. It is reasonable that TP-infpt is lower for New York than South Korea based on reports that South Korea has been more aggressive in testing and tracking contacts. Using the model values at day 58 as initial values, I ran the model again from that time point but with slight variations. It turns out that increasing TP-infpt rate by 10%, from 0.15 to 0.165 after day 58 has the same effect as increasing the number of tests by 50% to 37500 tests per day after day 58. The effect in both cases is to cut the number of cases at the peak by about 25% from 188K about day 120 to about 140K on or about day 108. These values should not be taken as predictions about absolute infected counts that might occur, but rather they suggest the relative importance of improving the efficacy of testing and increasing the number of tests. Figure 4 extends the simulation for a range of new parameter values that take effect after day 58. In the contour plot, starting from the lower left corner with TP-infpt rate of 0.15 and test per day at 25K, increasing TP rates have a bigger impact on decreasing the peak value of infected person. In the sample growth curve plot, change both TP-infpt rate and number of tests per day by 50% will dramatically change the growth curve. The contour plot shows maximum number of infected persons as TP-infpt rate and tests per day are increased. The increases are relative to the values using in the growth plot in Figure 3 which is similar to data observed in New York. The sample growth plot suggests that increasing TP rates and tests by a factor of 1.5 will start flattening the growth curve immediately.