|
Jun 2020
1:14pm, 6 Jun 2020
7,541 posts
|
jda
Yes it’s just nonsense.
|
|
Jun 2020
1:57pm, 6 Jun 2020
41,038 posts
|
Lip Gloss
That was my worry LMH that I hadn’t done it properly as too busy gagging and know it wasn’t for long enough. We had a leaflet to read first and it clearly stated one nostril.
I feel fine so I am going on the assumption that mine is a true negative
|
|
Jun 2020
3:37pm, 6 Jun 2020
498 posts
|
Non-runner
Cheers jda, I like a clear answer😊.
|
|
Jun 2020
4:11pm, 6 Jun 2020
11,213 posts
|
larkim
Local news in the NW of England is abuzz with "R is greater than 1" news (1.01 apparently). Whilst discussing in the house today, we wondered about the temporal context of R0. As I understand it, R0 is dimensionless, so a disease with a very, very low rate of transmission, but a 365 day duration, could have a R of 10, but that would only mean that on Jan 1 if 1 person had it, that by Dec 31 only 10 would have it (if it was only transmissible late in the disease cycle).
Wouldn't that be more comprehensible if R was expressed as R being something like "10 / 365 days". What would a similar expression say for Covid19? Is it currently nationally 0.7-0.9 / 4 days etc?
|
|
Jun 2020
4:46pm, 6 Jun 2020
7,545 posts
|
jda
Agree that a doubling/halving time is more immediately interpretable than an R number. R relates to around a 6 day change so you can work it out from that. Anything within a gnat's crotchet of 1 isn't really much to be immediately worried about to be honest in terms of a 2nd wave hitting us, though it does mean the disease isn't going away.
What is much more to be worried about, is that all of these estimates of R naturally lag any real changes by a week or two, so if the estimates are is rising and the latest estimate says it's gone up to about 1, that probably means that in reality it is above 1 already. At a minimum, it should be reported that the R estimate for late May is about 1.
1.01 is a silly level of precision of course, what it really means is that R is (oops I mean WAS!) as likely to be a little higher, as a little lower, than 1.
|
|
Jun 2020
5:01pm, 6 Jun 2020
11,214 posts
|
larkim
Yep, understood. Am I right in thinking that when you say R relates to a 6 day change, that’s a specific observation for this Covid infection? Or is that a generic feature of R?
Agree that when R is fundamentally an estimate, 1.01 is daft. With a relatively low level of infection (compared to earlier), one regions R could be heavily skewed by factors such as large infection rates within certain modelled communities (eg hospitals, care homes, social demographics with large households etc). What level of attempted precision do they go down to when segmenting the population into groups for R calcs? They can’t be (for example) suggesting that the behaviours of those in the NW of England are more likely to give rise to transmission (can they?), so any differential must be related to different population make-up and different estimates of the prevalence of infection in certain groups, mustn’t it?
|
|
Jun 2020
5:08pm, 6 Jun 2020
17,642 posts
|
EvilPixie
All these gatherings to protest are very worrying too
|
|
Jun 2020
5:17pm, 6 Jun 2020
7,548 posts
|
jda
That 6 day period specific to covid. It is an estimate of the average time taken from one person becoming infected, to their victims being infected. Eg for HIV it's much much longer.
In reality, we can only really estimate the doubling time directly from data, and turning this into R requires this reproductive time scale to be specified (possibly with some uncertainty). It's a bit silly to do it like this as the doubling time is more precisely constrained than either R or the reproductive time scale are, and it's the doubling time that really matters most. But R matters for policy interventions as they tend to change R by a factor.
Demographic influences, employment, housing, behaviour, infrastructure, just who happens to be infected at that particular time can all vary regionally. I think most of these estimates are aggregated at the regional level cos that is the level of precision afforded by the data.
|
|
Jun 2020
5:31pm, 6 Jun 2020
2,366 posts
|
Canute
Regarding Prof Karl Friston, I will give a much more long-winded answer than jda. I start with a declaration of possible conflict of interest. Karl’s first job as a research scientist was as a research assistant in a major brain imaging project that I led in the 1980’s. The questions that I asked him launched his subsequent stellar career in brain science. The goal of our project was to map patterns of brain function in schizophrenia. One of our challenges was developing a feasible way to analyse complex images of brain function. Karl came up with the mathematical solutions that led to the successful outcome of our specific project. Perhaps more importantly, he provided the foundation of modern brain imaging. Right from the beginning Karl’s work was controversial and he weathered much criticism. Nonetheless within a few years the majority of the brain imaging community world-wide was employing the analysis procedures developed by Karl, or procedures derived from his work.
I continued with my research into brain function and mental symptoms, while Karl tackled much more ambitious wide-ranging questions about the workings of the brain. He has now applied some of the mathematical modelling skills to the CV-19 epidemic. His approach is based on models that differ greatly in nature from more conventional mathematical models employed by epidemiologists. Not surprisingly, conventional epidemic modellers regard him with great suspicion. I have not followed the details of his current model and would not wish to defend it.
However I am concerned about conventional epidemic modelling of the CV-19 pandemic. There are computational procedures such as data assimilation that allow adjustment of the models so that they fit the current data well, but I am very dubious about their value for prediction beyond the very short term future. I am concerned that they do not deal adequately with the complexity of the mechanism of transmission of this virus. There are still some very puzzling features about the transmission of this virus. I would not neither defend nor dismiss Karl’s ideas.
|
|
Jun 2020
5:54pm, 6 Jun 2020
11,215 posts
|
larkim
Thanks jda. So for a simple presentation, r=1 means in 6 days time 1 more will be infected, and so on. R=2 would equate to 2 in 6 days, 4 in 12, 8 in 18, 16 in 24, 32 in 30. Add '000 to the end of the numbers and the scaling is immediately obvious.
That 6 day period surely should be more commonly referenced in the discussions about it? It gives a really important dimension when talking about the speed of response necessary, because all people are hearing about is R, and because of our over-simplistic understanding of it it either adds to the level of fear or complacency.
|
Coronavirus discussion thread
Feb 2024
