Read on Twitter
1/ After weeks of steady declines, US coronavirus case rates increased this week, with several states seeing record-high infection rates that taxed hospital and ICU capacity.
Closer to home, the Ventura County case load is growing along an exponential curve, not a linear one.
Closer to home, the Ventura County case load is growing along an exponential curve, not a linear one.
2/ Mike Bass ( @n7wlc) is a computational biologist with more than 25 years’ experience working with large data sets to pick out trends. As a fellow data geek, I’m grateful for Mike’s statistical chops in helping understand where we’ve been and where we might be heading.
3/ Awhile back Mike noticed something odd in the county’s numbers. Initially Mike had been plotting virus case data using a linear regression – one where the data progresses along a straight line.
4/ At first the actual number of cases more or less matched the linear trendline, or at least fell within the 95% prediction interval. This was during the first few weeks of the initial stay-safe-at-home orders.
5/ Then, starting in late April, county case numbers started rising fast. Now, the linear model no longer fit well.
6/ After multiple days of seeing numbers outside the prediction interval, Mike reran the model using an exponential regression – one where values increase along a hockey-stick curve.
7/ The exponential model fit much better -- and because of that, the prediction interval is much tighter. It’s not an exact fit, but the actual case numbers run much closer to the exponential curve than the linear one.
8/ Linear and exponential models matter because they predict how well our healthcare system can cope with the virus. When cases follow a linear curve, hospitals can keep up with a constant flow of new patients.
9/ In contrast, if exponential growth continues long enough, it will exhaust healthcare resources. Also, the longer exponential growth goes on, the faster rates will increase.
10/ Only a small percentage of coronavirus patients need hospitalization — but some will, and that makes the rate of increase in case rates important.
11/ As things stand now, Ventura County’s exponential growth is a cause for concern but not alarm.
12/ Only 43 of the county’s 585 active coronavirus cases are currently hospitalized, according to the state Department of Public Health . That’s not much of a burden for the county’s 1,353 licensed hospital beds. https://public.tableau.com/views/COVID-19PublicDashboard/Covid-19Hospitals?:embed=y&:display_count=no&:showVizHome=no
13/ The concerns here are twofold. First, we’ve been on exponential curve since late April, one that shows no sign of abating. If case growth continues at this pace, eventually it will put a burden on the healthcare system.
14/ Second, the county’s public-health officials aren’t talking about exponential growth. This week Mike shared his findings with supervisor Linda Parks, who passed the data along to the county public-health office. The unsatisfying response was that rates have “leveled off.”
15/ At a county supervisors’ meeting later this week, Ms. Parks asked county public health office Dr. Robert Levin if we are flattening the curve. He said he believes we are.
16/ County and state statistics don’t support the idea that new cases have leveled off. It’s true hospitalizations have stabilized after a sharp increase a couple of weeks ago, which is a good thing, and currently there is ample hospital capacity.
17/ Over time, though, continued exponential growth in cases could burden that capacity.
18/ As noted last week, the increase in cases cannot be explained by more testing alone. Testing is up, but so are hospitalizations, and that metric isn’t influenced by testing.
19/ Statewide, case growth fell between linear and exponential models for the past several weeks – but increases this week suggest they are following more of an exponential path.
20/ This suggests two possibilities: First, cases might continue exponentially from now on, as they have in Ventura County since late April.
21/ Second, state cases might actually be growing linearly, but using a two-line model, one shallower and one steeper, with more recent growth along the steeper path. At least for now, neither the single-line nor exponential models are a perfect fit to describe statewide trends.
22/ Nationally, cases grew at a faster rate, driven by sharp rises in several states, especially in the South. One positive note: Deaths increased at a slower pace than the previous week.
23/ For the week ending June 19, nearly 172,000 more Americans were confirmed positive for the novel coronavirus, and 4,443 died. These are respectively 8% and 4% increases over week-ago cumulative totals.
24/ In California, 24,203 residents tested positive and 448 died. These numbers went up more than twice as fast as the national totals, respectively at 17% and 9%.
25/ The rate of new cases per day in California has been on a steady upward climb since March. The rates are smoothed somewhat by averaging changes over 14 days, but the overall trend is one of steady upward growth.
