As we get ready for the new modelling data on COVID-19, let's take a look at how we've done compared to the last model, and reasons what actually happened was so different from what actually happened. Let's go! #covid19nfld

The modelling looked at hospitalization, so here's where it predicted we'd be now under current and worst case (yellow box around today).

The worst case scenario had us with 140 people in hospital right now. Even the "current" scenario had us with 25-30 in hospital #covid19nfld

So what actually happened? As of yesterday there were just 5 people in hospital, 20% of what the model predicted, and instead of slow and steady growth, here's what happened to our hospitalizations, it's been going down #covid19nfld

So what happened? The model assumed that each person who got COVID-19 would pass it on to 1.5 people. As of last week, each person was actually passing it on to 0.4, so instead of virus spreading, it's presence in NL has been shrinking #covid19nfld

Why was the model so far off what actually happened? A number of factors. One is that the model wasn't great to start with Dr. Rahman and the health minister said so at the time (remember his fuzzy crystal ball comment). The funeral home cluster made it harder #covid19nfld

A model shows what *could* happen, so people may have just been better at physical distancing, handwashing, they listened to the message from health officials and so changed the course of the disease (for the better) #covid19nfld

The health minister said the model would be used for planning, so the danger of not having great modelling data is that if they put their efforts into preparing for growing numbers, when they actually shrunk, they're now not prepared with a plan to ease restrictions #covid19nfld

So if the data isn't very good, should it be shared it with the public? Environment Canada does this all the time with modelling. Their computers say a storm will drop 40cm of snow 10 days from now, they don't release that info because they know it's not accurate #covid19nfld

So will we see very different modelling results today? Absolutely. Will it be better? It should because it has more data to go off on. Will it tell you for sure what's going to happen? Almost certainly not #covid19nfld