This page displays plots tracking COVID-19 data for Tompkins County, New York State, and the Finger Lakes region. I look at the number of tests performed and the number of confirmed positives. The data are downloaded from the New York State Department of Health, and are available here.The plots are interactive, hovering the mouse over them reveals available options to modify them and get more information about the underlying data. I update the plots daily. NOTE: I changed the plots and added some more information on Aug 1, 2020.
Here is the daily number of confirmed cases. I show the data for Tompkins County (where I live) alone and for the Finger Lakes region: Tioga, Chemung, Schuyler, Seneca, Cayuga, Cortland, Broome, and Onondaga counties (excluding Tompkins). The bars are stacked on top of each other, so the total height reflects the numbers for the whole region.
Here is the daily number of tests performed.
It is important to understand that these data are an imperfect reflection of the actual number of cases in the population. There are relatively few individuals tested, so the raw number is obviously a (possibly large) underestimate of the total infection prevalence. To see this, I plot the daily per capita number of tests (population data from here)
In addition, because of kit scarcity, tests are performed on individuals who are more likely to be positive than a random person. For example, tested people exhibit symptoms or think they have been exposed to an infected individual. This means that we cannot use the fraction of positives in this sample to directly extrapolate to the whole population. A larger number of tests would imply a sample that is more reflective of the total population (although biases can still exist, for example if the age distribution of the tested cohort is different from the population at large). A decrease in the fraction of people tested that turn out to be positive for SARS-Cov2 would also indicate an increasingly representative sample. It would also suggest that our ability to detect positive cases is not limited by the number of tests available. I follow this fraction, calculated from the daily number of positives, divided by the number of tests, in the last plot:
I added lines to the plot to smooth the daily fluctuations. They reflect seven day running averages.
Next, I plot the number of people hospitalized with COVID-19 in Tompkins county (data available here).
Finally, here is the numbe rof active COVID-19 cases in Tompkins county (from the same source as above).
If the daily number of confirmed cases stays small and the fraction of tested individuals that come up positive falls, while the number of daily tests increases, we can start to hope that the infection spread is being stopped, at least locally. However, making these judgments based on such data is tricky, so if any of the conditions I mentioned are not met, I would be doubtful that things are getting better. Ultimately, wide-spread reliable serological tests will be necessary to gain a more accurate picture of infection dynamics.
The R code I used to pull data and make the plots is on my GitHub page.