Watching news reports about COVID-19, Economics Professor Carsten Lange noticed that many of the state and local briefings focused on daily COVID-19 statistics such as positive test results and the number of deaths. In his experience as a data scientist, time series data provides a better understanding.
“Looking only at snapshots of COVID-19 cases and not at the trend over time gives the wrong picture,” Lange noted. “It’s active cases that reflect the true eminent danger of the pandemic. The daily growth rate of active cases reflects the short-term success or failure of managing the pandemic.”
To give the general public and local governments a more effective overview about the current state of the pandemic in California, Lange created a database that tracks predicted active cases for each of the 20 most populous counties, divided by north and south, using data collected by the Los Angeles Times.
“Active cases are the source of new infections, and the higher the number of active cases, the greater the chances of new infections,” said Lange. “Even small percentage increases will lead to catastrophic increases in active COVID-19 cases over a very short time. For example, a daily increase of 5% will double the active cases after only two weeks. In order to decrease the active cases to a sustainable level, it is not enough to reach a zero-percentage change. Instead, a negative double-digit percentage change over several days is needed.
“At the state level, the growth rate of active cases on May 10 was 2%, which means active cases are increasing instead of declining,” he said. “The trend in Los Angeles County is 0% growth in active cases, indicating that active cases are staying at the current level instead of improving. On the other hand, right now one of the best counties in California is Sacramento County which has a negative 4% growth rate in Active Cases.”
A professor of international economics, Lange specializes in data science, machine learning and geographic information systems. His COVID-19 model is based on the assumption that it takes an average of 14 days to recover from the virus infection. Consequently, active cases are cases less than 14 days old. The model is not intended to give advice to state or local governments, which should be done by epidemiologists, who use sophisticated models to predict and advise.