5 ms·

> while it's only effective if a majority of the population uses it? 60% is the minimum number the experts are saying must use the app before it's effective.

by s_dev 6y ago

cwhiz 6y ago
Source? I have heard much higher numbers, as high as 95%.
- jpollock 6y ago
  If (1-1/R0) of the population is using the app, it should bring the R0 below 1. With covid-19, if the R0 is 3, then being able to track 2/3rds of infections should _hopefully_ bring the R0 below 1. At least that's how I've been reasoning it out.
  - user5994461 6y ago
    The last estimates put R0 around 6.
    - chimprich 6y ago
      But do the best estimates? As far as I can tell estimates vary a lot, but an R0 of 6 would be a significant outlier. But the current R value in the UK under the lockdown is estimated to be somewhere between 0.5 and 1. A contact-tracing app doesn't need to be perfect; any reduction in R would allow for either a swifter reduction in cases or a liberalisation of lockdown measures.
      - user5994461 6y ago
        The virus is still spreading in every location in spite of the lockdowns. Don't believe R is below 1 even with the lockdown. The virus is extremely infectious. Random sampling in NYC was hinting there were 1-2 millions of people infected and that's not the only city like that.
        chimprich 6y ago
        > The virus is still spreading in every location in spite of the lockdowns. Don't believe R is below 1 even with the lockdown. The virus still spreads as long as R is over 0; you still get new infections if there's any spread. However, an R of 1 is the boundary for exponential spread. If infections are not spreading exponentially it has to be below 1 by definition (as I understand it). In the UK death rates and hospital admissions have been falling for some time, indicating R is currently well below 1. Neil Ferguson, one of the UK's top infectious disease modellers estimated the current R value at between 0.6 and 0.7 a couple of days ago: https://unherd.com/thepost/imperials-prof-neil-ferguson-responds-to-the-swedish-critique/ https://unherd.com/thepost/imperials-prof-neil-ferguson-resp... .
  - AaronFriel 6y ago
    That's assuming perfect compliance rates, right? I cannot estimate what the compliance rate for self-quarantine will be, but if we look at a few rates we get different numbers: 50% compliance? We already lost, too many people will ignore the app and we can't get to 2/3 even if 100% of the population installs. compliance rate | necessary install rate <67% failure (R_t stays >1) 67% 100% 80% 83% 90% 74% 95% 70% 100% 67% If as another person replied the R0 is around 6, then: compliance rate | necessary install rate 67% failure (R_t > 1) 80% failure (R_t > 1) 90% 93% 95% 88% 100% 83% These are... not great numbers for trying to deviate from what Apple and Google will push as a built-in app. Even assuming >= 90% of the country has a smartphone, you need a huge fraction of the population to participate.
    - icelancer 6y ago
      You are assuming only a single input, which is this app. R_t can be - and is - reduced through a multitude of methods, including but not limited to, closing of large public gatherings, high compliance of mask-wearing, increased disinfecting protocols, etc.
      - AaronFriel 6y ago
        That's a really good point! If we can reduce R_t with widespread social distancing, wearing masks, washing hands etc. down to <= 2 we are in much better place and this app could seal the deal and we could just about eradicate the disease. And even if we "only" drive R_t to almost 1 we can at least buy a lot of time for our healthcare workers, which is valuable. Either way, I didn't mean to be that pessimistic. I think an ensemble approach is necessary. For now, general (easy to follow) shelter in place rules, followed by test and trace using these apps, and then targeted relaxation of the rules as we discover what we can reopen without causing a spike in R_t.

> while it's only effective if a majority of the population uses it? 60% is the minimum number the experts are saying must use the app before it's effective.

by s_dev 6y ago

> while it's only effective if a majority of the population uses it? 60% is the minimum number the experts are saying must use the app before it's effective.

cwhiz 6y ago

Source? I have heard much higher numbers, as high as 95%.

jpollock 6y ago
If (1-1/R0) of the population is using the app, it should bring the R0 below 1. With covid-19, if the R0 is 3, then being able to track 2/3rds of infections should _hopefully_ bring the R0 below 1. At least that's how I've been reasoning it out.
- user5994461 6y ago
  The last estimates put R0 around 6.
  - chimprich 6y ago
    But do the best estimates? As far as I can tell estimates vary a lot, but an R0 of 6 would be a significant outlier. But the current R value in the UK under the lockdown is estimated to be somewhere between 0.5 and 1. A contact-tracing app doesn't need to be perfect; any reduction in R would allow for either a swifter reduction in cases or a liberalisation of lockdown measures.
    - user5994461 6y ago
      The virus is still spreading in every location in spite of the lockdowns. Don't believe R is below 1 even with the lockdown. The virus is extremely infectious. Random sampling in NYC was hinting there were 1-2 millions of people infected and that's not the only city like that.
      - chimprich 6y ago
        > The virus is still spreading in every location in spite of the lockdowns. Don't believe R is below 1 even with the lockdown. The virus still spreads as long as R is over 0; you still get new infections if there's any spread. However, an R of 1 is the boundary for exponential spread. If infections are not spreading exponentially it has to be below 1 by definition (as I understand it). In the UK death rates and hospital admissions have been falling for some time, indicating R is currently well below 1. Neil Ferguson, one of the UK's top infectious disease modellers estimated the current R value at between 0.6 and 0.7 a couple of days ago: https://unherd.com/thepost/imperials-prof-neil-ferguson-responds-to-the-swedish-critique/ https://unherd.com/thepost/imperials-prof-neil-ferguson-resp... .
- AaronFriel 6y ago
  That's assuming perfect compliance rates, right? I cannot estimate what the compliance rate for self-quarantine will be, but if we look at a few rates we get different numbers: 50% compliance? We already lost, too many people will ignore the app and we can't get to 2/3 even if 100% of the population installs. compliance rate | necessary install rate <67% failure (R_t stays >1) 67% 100% 80% 83% 90% 74% 95% 70% 100% 67% If as another person replied the R0 is around 6, then: compliance rate | necessary install rate 67% failure (R_t > 1) 80% failure (R_t > 1) 90% 93% 95% 88% 100% 83% These are... not great numbers for trying to deviate from what Apple and Google will push as a built-in app. Even assuming >= 90% of the country has a smartphone, you need a huge fraction of the population to participate.
  - icelancer 6y ago
    You are assuming only a single input, which is this app. R_t can be - and is - reduced through a multitude of methods, including but not limited to, closing of large public gatherings, high compliance of mask-wearing, increased disinfecting protocols, etc.
    - AaronFriel 6y ago
      That's a really good point! If we can reduce R_t with widespread social distancing, wearing masks, washing hands etc. down to <= 2 we are in much better place and this app could seal the deal and we could just about eradicate the disease. And even if we "only" drive R_t to almost 1 we can at least buy a lot of time for our healthcare workers, which is valuable. Either way, I didn't mean to be that pessimistic. I think an ensemble approach is necessary. For now, general (easy to follow) shelter in place rules, followed by test and trace using these apps, and then targeted relaxation of the rules as we discover what we can reopen without causing a spike in R_t.