Uses testing numbers and compares versus actual cases to adjust historic case counts for what the likely actual case counts were. They estimate in March only 8 percent of cases were detected by testing due to limited supply of test kits. They now estimate daily cases eclipsed 250,000 per day in the third week in March. That estimate would seem to match recent CDC estimates for total historic cases in the US. Based on the new estimates the total cases in the US based on their has flattened out since the third week of May at around 100,000 cases a day. If true that is a good result and would have the US on curves in other countries. They also conclude daily reported cases are rising due to greater testing so the actual cases are moving closer to the estimated cases each day but there are much fewer untested cases due to available tests.