WUHAN VIRUS DAILY UPDATE FOR APRIL 4
The downward trends continue for rates of growth for both incidents and deaths. Indeed we have new low records for both since the peak on Mar 19. Also - California test results have finally come in. Likely to be a re-sampling or correction for California but it helped us achieve a daily test rate of 170,129 tests! We've had this 60k+ pending test number since Mar 26th and now that's finally gone down to 15,573 pending test results. My expectation is that this is a one-time bump from California so we won't know what our real daily rate is for another day or two but this is good.
Given a total of 1,639,380 total tests completed in America, our new total for incidents is 305,755 (+33840), down to an increase of 12.45% over the previous day (down -1.32% from yesterday). Deaths have totaled 8,314 (+1,352) which is an increase of 19.42% (-0.95% rate of growth) over the previous day. Since the peak in Mar 19, rates of growth for incidents has fallen by 39.19% and for deaths has fallen by 23.44%.
If this decline curve holds, tomorrow we'll have a rate of growth for incidents under 10% for the first time since we've had data. THAT WOULD BE HUGE. We would see rate of growth for deaths drop below 10% by April 7 if current rolling averages sustain their velocity. HOWEVER - this is UNLIKELY because it makes the same flawed assumptions for our downward trends that all those other models that the press was touting as "exponential everything!" made - which is that all the other variables are stable - which they are not. *IF* these projections hold and the rates drop like they have been then this thing is basically over next week. It would mean that the hotspots are beaten and that there aren't new hotspots coming around the rest of the country. This is incredibly unlikely to be the case. I'd bet big money against it, unfortunately. I'm reporting it because that's what the model shows - but I'm also trying to make it clear what my favorite statistician, George P. Box, famously said, "All models are wrong, some are useful."
As the virus spreads out across the rest of the country, each region is going to have its own growth curve which means our numbers in America are going to start going squirrely and since single figures for America as a whole start becoming pointless in identifying how the pandemic is behaving - we're just too large. That's also why the world-wide numbers are something I've always ignored because they're bogus and pointless even if accurate. You'd have a better predictive indicator if you used a dart board - fact. So that's why I'm trying to develop my regional model to identify and isolate hotspots and report them independently of areas that are contained and areas that are at-risk - (I'm assuming I'll be able to qualify these 3 categories - we'll see). Unfortunately my other work hasn't given me the time to make any progress on this since last weekend so I really can't say when this will be ready.
Until then - I'll continue to report on my current model until I determine that it is more misleading than helpful in terms of guidance. No matter what, however, our national and local policies MUST be to start getting widespread random testing as soon as possible. It's the only way to guide our policies about this pandemic in an intelligent manner that doesn't end up causing more harm than good.
You all keep your hygiene practices going, avoid crowds, yes masks do provide some positive protection when around others, and be extra careful to protect the elderly and at-risk from exposure of any kind. That's ultimately the only attitude that beats this.