Since the lockdown began, my 5-year-old daughter has been getting into Facetime on the iPad. It is wonderful to see her having long sessions with her grandma and her friends, sharing her latest imagined worlds, ideas, drawings and creations. (And invaluable for her busy parents trying to fit in some work during these crazy times…)
However, I kept dreading the time when the iPad would inevitably get dropped and break. I needed a way that my daughter could carry it around, setting it up anywhere she wanted to, without it falling over.
Off to the shed…
I found that I still had some aluminium angle lying around and plenty of pop rivets.
After a very pleasant afternoon, I ended up with this:
The handle is made from one of those crossed tie-bars that Ikea provide to make their shelving more rigid. I cut off the ends, bent them, and stuck them into a piece of bamboo cane. It even has 4 rubber feet taken from an old amplifier. Of course, I had to file and sand down all the sharp edges.
Here it is with iPad in situ…
It’s really working well. She carries it around with her, puts it down anywhere, tilts it to point at things. Job done!
During the week following my last post, the UK’s Covid-19 death rate was actually between 700 and 900, not between 900 and 1300 as my model predicted. (It is still very high of course.)
A possible reason emerged when the government changed their Covid-19 dashboard. It is now at https://coronavirus.data.gov.uk/. On the new site they explain that the daily cases previously published were often several days late in being reported. On the new dashboard, they now attach the cases to the date when the patient was actually tested. This means they regularly add new cases to dates already passed and published (and that therefore the numbers for the past 5 days should be considered incomplete).
If you’re interested, I re-ran the model using the new data. There is still a similar correlation (this time to 8 days previously, not 7). But, because the last 5 days are not complete, it’s not possible to attempt a prediction for further ahead than 3 days.
As you can see from the chart below, the prediction for the next three days is that the death rate will remain about the same. We will see. Hopefully, it will be much lower than this. Of course, if the UK’s testing regime were to improve drastically (as everyone is calling for) the case-rate would increase significantly, without a corresponding increase in fatalities.
Looking at the UK Covid-19 numbers, the daily death rate seems to lag the daily cases, which makes total sense. I wondered what the lag was and generated the chart below which shows the best correlation between the two is when the delay is 7 days. In other words, the ups and downs in both the death-rate and the case-rate match each other best when the case-rate is shifted back by 7 days.
Then I looked at the relationship between the death rate on any particular day and the new cases from 7 days before. There seems to be a clear linear fit:
So, roughly speaking, the daily death rate can be estimated by taking the case-rate from 7 days before and multiplying by 0.23. Using the relationship above gives the following prediction for the next week:
Conclusion: If this is simplistic model turns out to be close to the truth, we should brace ourselves for a death-date between 900 and 1300 for the next week.
Caveat: I am no expert on this. I just got curious about the numbers. Those modelling the pandemic will be creating much better predictions.
P.S. Log scale: Below is the same chart on a ‘log’ scale (which emphasises the multiplication rate rather than the increase rate). You can see the hopeful “flattening of the curve”. But don’t forget this is not the accumulation — we are adding this number of deaths each day so of course we are desperate for it to come down fast not just stay flat.