Geekzone: technology news, blogs, forums
Guest
Welcome Guest.
You haven't logged in yet. If you don't have an account you can register now.


Filter this topic showing only the reply marked as answer View this topic in a long page with up to 500 replies per page Create new topic
1 | ... | 530 | 531 | 532 | 533 | 534 | 535 | 536 | 537 | 538 | 539 | 540 | 541 | 542 | 543 | 544 | 545 | 546 | 547 | 548 | 549 | 550 | ... | 773
3741 posts

Uber Geek

Lifetime subscriber

  #2464560 18-Apr-2020 09:19
Send private message quote this post

What does 96% accurate mean? Does it mean it will find 96% of cases (I.e. a 4% false negative rate)? Or that there is a 4% false positive rate? Or some combination of false positives and false negatives that adds up to 4%?



Mad Scientist
22365 posts

Uber Geek

Trusted
Lifetime subscriber

  #2464563 18-Apr-2020 09:30
Send private message quote this post

frankv: What does 96% accurate mean? Does it mean it will find 96% of cases (I.e. a 4% false negative rate)? Or that there is a 4% false positive rate? Or some combination of false positives and false negatives that adds up to 4%?

 

more importantly ... is it 96% accurate. i'm not saying it isn't but needs to be checked

 

edit - referring to these - (there are more countries involves, basically every european country i have googled reports the same)

 

https://www.newsweek.com/uk-says-millions-coronavirus-test-kits-bought-china-unreliable-most-patients-1496506

 

https://www.scmp.com/news/china/society/article/3077169/spanish-capital-ditches-unreliable-chinese-coronavirus-test-kits

 

https://www.nationalreview.com/news/u-k-purchased-millions-of-unreliable-coronavirus-tests-from-china/

 

https://www.bloomberg.com/news/articles/2020-04-01/faulty-virus-tests-cloud-china-s-european-outreach-over-covid-19

 

https://www.reuters.com/article/us-health-coronavirus-china-testkits/china-clamps-down-on-coronavirus-test-kit-exports-after-accuracy-questioned-idUSKBN21J51S





Involuntary autocorrect in operation on mobile device. Apologies in advance.


 
 
 
 


10316 posts

Uber Geek


  #2464567 18-Apr-2020 09:47
Send private message quote this post

frankv: What does 96% accurate mean? Does it mean it will find 96% of cases (I.e. a 4% false negative rate)? Or that there is a 4% false positive rate? Or some combination of false positives and false negatives that adds up to 4%?

 

I believe that as sensitivity is increased then selectivity reduces - so they can make more sensitive tests that are less likely to produce a false negative from not detecting antibodies that were in the sample, but that increases the rate of false positives as the test isn't specific enough to show only C-19 antibodies.  

 

That paper I linked to above, they used some negative and positive samples as controls (from PCR diagnosed patient blood samples), with the conclusion:

 

A combination of both data sources provides us with a combined sensitivity of 80.3% (95 CI 72.1-87.0%) and a specificity of 99.5% (95 CI 98.3-99.9%)

 

So the test kits used were not very sensitive - but quite selective.

 

As I've said a couple of times - something weird is going on, they did get 1.5% positives (so still possible with 98.3% selectivity for those to all be false positives).
If not, then the data conflicts with what we see in populations like NZ and Aus, Korea, Taiwan etc - at least it conflicts unless there are people getting asymptomatically infected but amongst those asymptomatic cases, very few are transmitting the infection to others.  If not - then why not?


10316 posts

Uber Geek


  #2464579 18-Apr-2020 10:35
Send private message quote this post

Batman:

 

more importantly ... is it 96% accurate

 

edit - referring to these - (there are more countries involves, basically every european country i have googled reports the same)

 

 

Haven't looked at all of those, but are they the same kits?  I think in Denmark they did a population survey, but the kits only had selectivity (US FDA interim approved) stating selectivity was (IIRC) 96.4%.  They found from that IIRC 1.7% of the population was "positive" - but that's well within the known possible false positive rate of 3.6% - and even that 3.6% figure has a margin of error - it could be more.

 

I think the ones the UK bought were worse - and thus worse than completely useless for intended purpose (home testing to assume immunity).  So if you assume you're immune because of a flawed test and work as a nurse with C-19 patient contact with reduced PPE, then you get infected and possibly die - as well as spreading the infection that you didn't think you could get.  It's more dangerous than no testing at all.

 

I'd trust our own Souxsie Wiles as someone to listen to - without any agenda apart from to ensure the best public health outcome.


Devastation by stupidity
12243 posts

Uber Geek

Lifetime subscriber

  #2464596 18-Apr-2020 10:52
Send private message quote this post

Fred99:

 

As I've said a couple of times - something weird is going on, they did get 1.5% positives (so still possible with 98.3% selectivity for those to all be false positives).
If not, then the data conflicts with what we see in populations like NZ and Aus, Korea, Taiwan etc - at least it conflicts unless there are people getting asymptomatically infected but amongst those asymptomatic cases, very few are transmitting the infection to others.  If not - then why not?

 

 

The science behind this is frankly over my head, but is it not possible that there could be slight genetic variations in the virus around the world, accounting for regional differences in the way populations respond to it? Maybe our part of the world is just lucky.

 

 

 

 





I don't think there is ever a bad time to talk about how absurd war is, how old men make decisions and young people die. - George Clooney
 


934 posts

Ultimate Geek

Lifetime subscriber

  #2464598 18-Apr-2020 10:56
Send private message quote this post

I know that the WHO has minimal credibility at this point, but they are warning against assuming antibody tests show evidence of immunity.

 

https://www.cnbc.com/2020/04/17/who-issues-warning-on-coronavirus-testing-theres-no-evidence-antibody-tests-show-immunity.html

 

 

 

My take is that there is yet to be a scientific consensus that immunity from having covid-19 lasts a useful length of time. As such, we should not divert effort to antibody testing from other means of fighting covid-19 until a science is a scientific consensus is formed. We have the luxury of letting other countries be the test subjects for antibody testing.




Mad Scientist
22365 posts

Uber Geek

Trusted
Lifetime subscriber

  #2464600 18-Apr-2020 11:02
Send private message quote this post

Fred99:

 

Batman:

 

more importantly ... is it 96% accurate

 

edit - referring to these - (there are more countries involves, basically every european country i have googled reports the same)

 

 

Haven't looked at all of those, but are they the same kits?  I think in Denmark they did a population survey, but the kits only had selectivity (US FDA interim approved) stating selectivity was (IIRC) 96.4%.  They found from that IIRC 1.7% of the population was "positive" - but that's well within the known possible false positive rate of 3.6% - and even that 3.6% figure has a margin of error - it could be more.

 

I think the ones the UK bought were worse - and thus worse than completely useless for intended purpose (home testing to assume immunity).  So if you assume you're immune because of a flawed test and work as a nurse with C-19 patient contact with reduced PPE, then you get infected and possibly die - as well as spreading the infection that you didn't think you could get.  It's more dangerous than no testing at all.

 

I'd trust our own Souxsie Wiles as someone to listen to - without any agenda apart from to ensure the best public health outcome.

 

 

definitely not the same kits, those were diagnostic kits, while the discussion now is about antibody test kits.

 

I'm just saying that it appears that out of desperation, buyers are skipping regulatory standards due to the rapid evolution of current conditons but at some expense of quality control





Involuntary autocorrect in operation on mobile device. Apologies in advance.


 
 
 
 


5918 posts

Uber Geek

Trusted
DR
Lifetime subscriber

  #2464601 18-Apr-2020 11:05
Send private message quote this post

Problems with 'nursing homes' in the US ...

 

 

 

The Washington Post - Hundreds of nursing homes in areas with outbreaks have repeatedly violated infection control rules

 

today

 


Forty percent of more than 650 nursing homes nationwide with publicly reported cases of the coronavirus have been cited more than once by inspectors in recent years for violating federal standards meant to control the spread of infections, according to a Washington Post analysis.

 

Since 2016, the nursing homes accrued hundreds of deficiencies for unsafe conditions that can trigger the spread of flu, pneumonia, urinary tract infections and skin diseases.

 

Dozens were flagged by inspectors only months before the coronavirus pandemic struck the United States.

 

Among the facilities with infection-control infractions: the Pleasant View Nursing Home in Mount Airy, Md., where 24 people had died as of Thursday; the Canterbury Rehabilitation & Healthcare Center near Richmond, with 49 deaths as of Thursday; and the Brighton Rehabilitation and Wellness Center in southwestern Pennsylvania, where officials have warned that all 750 residents and staff members could be infected. ...

 

 

 

 

EDIT

 

 

 

The New York Times - ‘They’re Death Pits’: Virus Claims at Least 7,000 Lives in U.S. Nursing Homes

 

today

 

 

More than six weeks after the first coronavirus deaths in a nursing home, outbreaks unfold across the country. 

 

About a fifth of U.S. virus deaths are linked to nursing facilities.

 

... more than 36,500 residents and employees across the nation have contracted [COVID-19]. ...

 





Sideface


739 posts

Ultimate Geek

Subscriber

  #2464605 18-Apr-2020 11:18
Send private message quote this post

tdgeek:

 

DS248: Incidentally I will post a plot in the data thread a bit later showing that the 'drop' in cases to 8 yesterday is an illusion.  In reality our new local cases confirmed per day has remained ~flat over the last four days (at ~12/day).  In fact at 13, yesterday was the largest number over the last four days. ... 

 

Can you explain. I see the 8 cases advised today. Great. All are related to existing cases. ...

 

Your comment?

 

Consider five significant dates:  Date infected; Date of symptom onset (if any, but normally 5-6 days after infection); Date tested - date on which sample is taken from a person by a District Health Board community testing centre or medical practice etc (test organisation). Sample then sent to a lab for testing; Date of report - date that the test lab reports the result, which generally seems to be 1 - 2 days after the date tested (though anecdotally sometimes can be longer); Press release date - the date on which that result is included in the MOH daily briefing. Could add date of infectiousness to that but not needed in this explanation.

 

The trend of new cases vs date infected would be best for assessing prevalence of COVID-19 vs time but that is often unknown so is not possible.  The next best option would be to plot new cases vs onset of symptoms.  Hong Kong does provide that info (for those symptomatic at time of testing) and Taiwan does for some cases but I am not aware of other countries that do so, at least publicly.  Probably there are but certainly not NZ.  From the MOH publicly released individual case data, the best we can do is look at the trend of new cases vs 'date of report' which I assume to be effectively date of confirmation that a person is COVID-19 positive.  It would be helpful, with zero privacy issues and very little effort if MOH at least included 'date tested'.  But 'date of report' at least is a reasonable proxy for that since in most cases the results (if positive) seem to be available within a day or two of testing. 

 

'Press release date' is a poor substitute for 'date tested'/'date of report', at least for the MOH data here in NZ as it is very sensitive to when results come into MOH and when they tally them.  In the MOH briefing the 'new' cases reported (but called 'change in last 24 hours on their website', which seems more accurate) typically include cases reported over the day or two before the briefing (usually most the day before), and sometimes a few from earlier dates.  And recently often one or two reported earlier in the day of the briefing.

 

Unusually, the data provided in the MOH briefing on 16 Apr, included 7 cases (1 imported, 6 other) that had a 'date of report' of 16 Apr; that is cases that had been confirmed earlier in the day of the briefing.

 

The press briefing yesterday (17th) added a further 8 (1 imported, 7 other) to the total with a 'date of report' of 16 Apr.  There were also a few +1 & -1 for other dates but they cancelled out.  And there were none for the current day (17th), hence the headline 8 'new' cases at the press briefing.  Net just 8 added to the 16 Apr report date.  Which on the face of it made our trend look very good, heading steeply down to near zero.  But as I noted the headline 8 new cases yesterday is an illusion - they constitute only ~half the cases reported by test labs on 16 April (reflecting samples taken ~2 days earlier). 

 

As per the plot below, the number of new cases vs 'Date of report' has been reasonably flat for the last four days.  In fact an uptick in new cases on the 16th (no data yet for the 17th).  Definitely not a sharp downwards trend as the headline press release data suggested.  NB: plot below is similar to the one I previously posted but I have included total new cases as well as local (imported being the gap between the two curves).

 

It will be interesting to see today's figures.  In this instance, no cases for 17 Apr have been publicly released yet so today's figure will be 'ground up'.  But also need to be aware that tomorrow's press release may include some for the 17th so the figure for the 17th in the individual case data they release today may not be complete (equally the figure for the 16th may still be subject to change).

 

 

 

 

 

 

 

 

 


10316 posts

Uber Geek


  #2464608 18-Apr-2020 11:27
Send private message quote this post

Batman:

 

definitely not the same kits, those were diagnostic kits, while the discussion now is about antibody test kits.

 

I'm just saying that it appears that out of desperation, buyers are skipping regulatory standards due to the rapid evolution of current conditons but at some expense of quality control

 

 

A quick look at your links - they were "quick" antibody test kits.  The ones that the Spanish bought - according to the SCMP used a nasal swab rather than blood sample - but still tested for antibodies or antigens, not viral RNA sequences - as the PCR tests we use here do. 

 

But yeah - there's desperation and some big mistakes being made.  The US churned out PCR test kits that were useless, they knew they were useless, there was an alternative (WHO), but I guess MAGAness and/or holding too dearly on to some notion that "American Exceptionalism" would rescue them has doomed them - a country with  5% of the world population has 32% of world confirmed cases and 25% of world recorded deaths.  They still haven't got it anywhere near under control and they're wanting to come out of lockdown.


10316 posts

Uber Geek


  #2464620 18-Apr-2020 11:44
Send private message quote this post

Rikkitic:

 

The science behind this is frankly over my head, but is it not possible that there could be slight genetic variations in the virus around the world, accounting for regional differences in the way populations respond to it? Maybe our part of the world is just lucky.

 

 

It's possible that the different genomes may have different characteristics, but very hard to prove unless some very obvious difference in how the disease progresses comes to light - or the mutation they observe will cause a known difference in how the virus functions.

 

By 10 April, nextstrain had logged over 3,000 genomes from samples from all over the world.  These were only the mutations analysed and data uploaded to the public domain. So they can trace disease spread using this data, but it's a massive task to correlate data on how severe cases are, isolate the genome, and conclude that there are milder and more severe strains.  Most countries with many cases have many strains as the disease came into the country from different places.

 

According to some theories on viral epidemic evolution and observation, some viruses become less severe but more contagious.  I don't think there's any evidence this has happened yet with C19.  It's a "keep your fingers crossed" situation IMO. 


2929 posts

Uber Geek

Trusted
Lifetime subscriber

  #2464626 18-Apr-2020 11:56
Send private message quote this post

Just wow.... If this man means we have to stay at level 4 for another month.... Lost for words...

 

 

 

https://www.stuff.co.nz/timaru-herald/news/121102617/man-with-covid19-arrested-after-shopping-at-timaru-supermarket





My views (except when I am looking out their windows) are not those of my employer.

739 posts

Ultimate Geek

Subscriber

  #2464627 18-Apr-2020 11:56
Send private message quote this post

I meant to add to my previous post that, based on the trends in the plot, it seem unlikely we will be in a sufficiently favourable position by Monday for a decision to drop to L3 two days later. 

 

As mentioned previously, I personally think a weeks extension would be best - allows time to see how our downward trend is progressing and for more data from the recently started targeted testing to come through.


1835 posts

Uber Geek

Trusted
Subscriber

  #2464644 18-Apr-2020 12:17
Send private message quote this post

DS248:


As mentioned previously, I personally think a weeks extension would be best - allows time to see how our downward trend is progressing and for more data from the recently started targeted testing to come through.


I think it's a bit daft to go off level 4 on Wednesday and then have 2 days of a working week and then a 3 day weekend ANZAC where everyone will tend to go a bit overboard with travel etc despite best intentions.

At the minimum defer until the Tuesday after ANZAC weekend. Gives us another 6 days without too much hassle







My thoughts are no longer my own and is probably representative of our media-controlled government




Mad Scientist
22365 posts

Uber Geek

Trusted
Lifetime subscriber

  #2464647 18-Apr-2020 12:19
Send private message quote this post

hairy1:

 

Just wow.... If this man means we have to stay at level 4 for another month.... Lost for words...

 

 

 

https://www.stuff.co.nz/timaru-herald/news/121102617/man-with-covid19-arrested-after-shopping-at-timaru-supermarket

 

 

Reminds me of the catch-phrase that is used ... "high trust environment" ... we've said repeated it doesn't work.

 

If we're serious about eradication it's time to quarantine every case and every incoming: passenger and crew and diplomat and animal.





Involuntary autocorrect in operation on mobile device. Apologies in advance.


1 | ... | 530 | 531 | 532 | 533 | 534 | 535 | 536 | 537 | 538 | 539 | 540 | 541 | 542 | 543 | 544 | 545 | 546 | 547 | 548 | 549 | 550 | ... | 773
Filter this topic showing only the reply marked as answer View this topic in a long page with up to 500 replies per page Create new topic



Twitter and LinkedIn »



Follow us to receive Twitter updates when new discussions are posted in our forums:



Follow us to receive Twitter updates when news items and blogs are posted in our frontpage:



Follow us to receive Twitter updates when tech item prices are listed in our price comparison site:





News »

Menulog change colours as parent company merges with Dutch food delivery service
Posted 2-Jul-2020 07:53


Techweek2020 goes digital to make it easier for Kiwis to connect and learn
Posted 2-Jul-2020 07:48


Catalyst Cloud launches new Solutions Hub to support their kiwi Partners and Customers
Posted 2-Jul-2020 07:44


Microsoft to help New Zealand job seekers acquire new digital skills needed for the COVID-19 economy
Posted 2-Jul-2020 07:41


Hewlett Packard Enterprise introduces new HPE GreenLake cloud services
Posted 24-Jun-2020 08:07


New cloud data protection services from Hewlett Packard Enterprise
Posted 24-Jun-2020 07:58


Hewlett Packard Enterprise unveils HPE Ezmeral, new software portfolio and brand
Posted 24-Jun-2020 07:10


Apple reveals new developer technologies to foster the next generation of apps
Posted 23-Jun-2020 15:30


Poly introduces solutions for Microsoft Teams Rooms
Posted 23-Jun-2020 15:14


Lenovo launches new ThinkPad P Series mobile workstations
Posted 23-Jun-2020 09:17


Lenovo brings Linux certification to ThinkPad and ThinkStation Workstation portfolio
Posted 23-Jun-2020 08:56


Apple introduces new features for iPhone iOS14 and iPadOS 14
Posted 23-Jun-2020 08:28


Apple announces Mac transition to Apple silicon
Posted 23-Jun-2020 08:18


OPPO A72 a top mid-tier smartphone
Posted 19-Jun-2020 18:02


D-Link A/NZ launches new smart AX1500 Wi-Fi 6 Router
Posted 19-Jun-2020 15:03



Geekzone Live »

Try automatic live updates from Geekzone directly in your browser, without refreshing the page, with Geekzone Live now.


Support Geekzone »

Our community of supporters help make Geekzone possible. Click the button below to join them.

Support Geezone on PressPatron



Are you subscribed to our RSS feed? You can download the latest headlines and summaries from our stories directly to your computer or smartphone by using a feed reader.

Alternatively, you can receive a daily email with Geekzone updates.