iom_dave

Random question, have you had a look at the gov.je website. I think they might have an actual epidemiologist. https://www.gov.je/Health/Coronavirus/pages/coronaviruscases.aspx They have an R - currently 1.4-1.7 And per here 10 people in hospital. Looking roughly, they seem to be a couple of weeks ahead. https://www.itv.com/news/channel/2020-03-12/live-updates-number-of-coronavirus-cases-in-the-channel-islands

One thing, i think the numbers they are comparing to Europe are out by a factor of 7: The current Isle of Man rate of infection per 100,000 is roughly 277. For context, in Europe, only Cyprus, Gibraltar, the UK, Netherlands and Spain are higher. 1 The number of cases in Europe are quoted in terms of per week (as far as I can tell), whereas the IOM is running at 277 per day while the UK is 755 per fortnight. https://qap.ecdc.europa.eu/public/extensions/COVID-19/COVID-19.html#global-overview-tab Really hoping I've got the wrong end of the stick somewhere.

Had a bit more think about the above, needs an engineering adjustment of x2, maybe 10 beds. Justification is that the IOM is a much smaller population, the UK will be benefiting from averaging between high and low areas. Models are only as good as their data inputs/calibration, each wave of covid in each country seemed to be unique so there are very limited data inputs - it turns into expert judgement. This is why you need an epidemiologist to be in charge of the modelling working with a mathematician to do the implementation. Main things are get vaccinated, be sensible with masks and try to give people space even if they aren't aware they need it.

Is that with or without mask?

Good to hear that it is heading the right direction. The final bit re 18 year olds is the STEP papers, though that is a bit specialist - does anyone support that in the Manx schools, I just remember it seeming like another level at the time! Getting a bit off topic...

I don't live in the IOM, but have friends and family there. Looking at your twitter page am really pleased that you are pushing Further Maths - I was the first to do Further Maths for 5 years at CRHS - would like to give credit to my Maths teacher Mr Horan as there was no timetabling for it but would have been better had some others had been encouraged too! Regarding the questions on a and b, your model set up is to only look at children so cannot really help on 2+2 issues. My feeling based on current trends on the ONS is that 50+ just aren't getting it anything like as much and the vaccines are working (see graphs below), this is also reflected in the more recent cuts of the manx data - it is actually quite a good tool though they really don't help themselves with failing to calc the diff on the front page each day! I would say that 2+2 high risk should isolate, but that is common sense. At an individual level, a high risk 2+2 person has no idea if they have good immunity and they need to consider whether a trip to a pub full of young people is worth it. I suspect they won't be going there anyway. The more interesting question I think is for the 15-30s and long covid. They should be strongly encouraged to keep safe, they have a whole lifetime of long covid side effects to live with and have already had to put up with 16 months of covid rubbish, one more month and they will be fully vaccinated, but most have had one jab so hopefully wont be too bad. https://www.ons.gov.uk/peoplepopulationandcommunity/healthandsocialcare/conditionsanddiseases/articles/coronaviruscovid19latestinsights/infections The IOM is showing similar low proport in the over 50s (admittedly I havent got the raw data to directly compare the uk and the iom) Most 50+ not vaccinated (possibly group 1 was in the date range) Most 50+ vaccinated

Yes, that is how it would would. I think a good question is why does the iom need a model. As a first order approximation, is max English hospital beds / 500 not approximate enough? Based on the below, that would be at approximately 5 beds a day. England is more crowded and has less vaccine uptake. These guys are far more qualified that anyone on here and are being subject to peer review by much more qualified people: https://assets.publishing.service.gov.uk/government/uploads/system/uploads/attachment_data/file/1001172/S1302_University_of_Warwick_Road_Map_Scenarios_and_Sensitivity_Step_4.2__6_July_2021__1_.pdf Occam's Razor...

Maybe not 100% true, but did you see the Euro final - that was crazy, put 3/4 of the iom in stadium and have a riot! The government over here shouldn't have relaxed the May restrictions until the 18 year olds had a chance to be 2+2

Not sure in reality that much is happening in the uk - there is still the tube being rammed everyday etc. The Isle of Man has a much lower density of population so it naturally has social distancing compared to cities.

That is quite a reply, I feel I'm asking a reasonable question, you seem to be applying the model incorrectly and failing to explain in layman's terms why that is acceptable. I understand that a lot of people are stressed at the moment. Does you model not approximate what I described above, which isn't the situation given the vaccination status? Btw, in the past I have used these sort of models for foot and mouth, my conclusion then that they are hugely sensitive to the underlying assumptions, hence the questioning around R0 and stratifying the population.

The reason that we are not querying R37=1.1 is that R0=3 is the cause of the explosion, I think that is first order issue with the model. The model needs R0 to be representative of transmission between member of the population, when 75% of the population is immune then you need to adjust R0 (or the model initial values). There is a huge difference between the two approaches. Going back to basic principles on R0 Current state of the iom, assuming R0=4 and one large population 1st gen A gets covid, meets B1, B2, B3 ad B4. B2-4 are vaccinated, only B1 gets covid. 2st gen B1 meets C1, C2, C3, C4... and so only one person if gen gets covid. Your model instead assumes with R=3 and none of the vaccinated meet the infected 1st gen A gets covid, meets B1, B2, B3. No one is vaccinated everyone gets covid, the number with covid explodes. The main people the infected (kids) interact with are their parents (particularly in the summer holidays), and they are vaccinated, this will hugely slow down the spread. The reason I suspect there has been an outbreak is that a few 2+2 people with covid have silently spread it amongst the iom population, no one really noticed for a few weeks until it started to infect enough to be picked up, then people got worried. Then everyone started to get tested and now there are loads being reported. Hopefully now people will start taking some sensible measures and the infection rate will drop. I wish they hadn't introduced 2+2 at the end of June as could have been avoided so more adults could have been vaccinated but I think the model is giving overly scary predictions. These models are hugely sensitive to R0 and the assumptions you make about the population mixing. I suspect that as the IOM has a higher % vaccine coverage than the UK it will do better than the UK.

Maybe my point re R0 isnt quite clear. R0 is for the whole population, but you have ignored 75% of the population by removing them. The initial conditions should have "recovered"=75%, this should greatly reduce the spreading at the start. Does the model implementation let you set the "recovered" initial condition? I suspect the "explosion" in numbers in the IOM in the last couple of weeks is due to pubs/clubs and 15-30s. This will probably slow down as it sounds like pubs are shutting and people are pinging. The IOM is small enough that a couple of "super spreading" events could drive the numbers. Kids are especially difficult to model as they alternate hugely depending on school holidays. Assuming a nuclear family of 4 with two kids on summer holidays mostly in their house. Kid A infects Kid B, end of infection as the parents are immunised. I kind of feel re kids that it is hard to justify as the benefit to them is low and there are so many people over 70 in the world haven't been jabbed.

So, just looking through the modelling approach from Ben, I think you are splitting out the population of the unvaccinated (Kids) from the vaccinated but then using R based on an unvaccinated population. I'm not quite sure how you set R0 (key unknown in these sort of models), is it the original R0? In simple terms it feels a bit as though the model is taking all the kids to the Calf of Man and then giving them Covid 1.0. Issues with this: 1. R0 is based on the population having no immunity, in this case 64k+5k should be immune, reducing the changes greatly of finding someone to pass it on to. This should reduce the R0 in the model starting conditions. 2. R0 for delta is much higher (maybe 6?) but I think the net effect (combined with 1) is probably something like 1.5 - similar to the R0 in the UK. The UK has more masks but less immunisation and is also much more densely populated - e.g. the tube. 3. Very few kids are getting covid - see stats below As an FYI, when I first heard about 2+2 in June I thought it was madness as vaccinated can carry. I still think it was a bad idea and they should have waited a month to allow under 30s to be vaccinated. Having looked at the vaccination stats and current infection stats I think it might be ok to badish (rather than the terrible forecast above). The kids dont seem to be getting it (the population in the model above) and about to go on summer holidays, the 15-30s are getting it, but as they have had a first mostly had at least one jab, they will have antibodies of some level and not end up in hospital - this is supported by the low number of hospital cases at the moment (though there is a 2 week lag on this). https://covid19.gov.im/general-information/covid-19-vaccination-statistics/ https://covid19.gov.im/general-information/latest-updates/