|
@@ -0,0 +1,125 @@
|
|
|
+# Flawed Data Driving the Narrative
|
|
|
+## Intro
|
|
|
+- Bayesian Networks - Smart data > Big data
|
|
|
+- appled in Law and Forensics
|
|
|
+- looking at data for 22 months
|
|
|
+- concern about how statistics are used to drive the narrative
|
|
|
+- lack of evidence for lockdowns and mandates
|
|
|
+- bias to work: even if as dangerous as claimed and measures as effective as claimed, no justification for lockdown measures and vaccine passports
|
|
|
+- don't need sophisticated statistical skills to understand that the threat and preventative measures are overstated and exaggerated
|
|
|
+- qualified people remain silent on this
|
|
|
+
|
|
|
+## Itinerary
|
|
|
+1. Censorship
|
|
|
+2. Data problems
|
|
|
+3. Vaccine efficacy/safety (all-cause mortality)
|
|
|
+
|
|
|
+## Specialization
|
|
|
+- Bayesian-causal modeling techniques
|
|
|
+- use publicly-available data
|
|
|
+- national differences in CFR/IFR
|
|
|
+- published in peer-review
|
|
|
+- called for random testing early (regret this)
|
|
|
+
|
|
|
+## Censorship
|
|
|
+### Initial perspective
|
|
|
+- Infection rates higher than predicted
|
|
|
+- Fatality rates lower than predicated
|
|
|
+- Disease less lethal than what was reported -> before effective treatment protocols were available
|
|
|
+- Extensive research into risk factors/symptoms
|
|
|
+- Incorporate data into causal model
|
|
|
+- Packaged as personalized symptom/risk assessment app (web/mobile)
|
|
|
+- Exposed flawed in reports about particular risk factors
|
|
|
+- Looked into ethnicity
|
|
|
+- Extensive work on symptom tracking apps
|
|
|
+- All work was well-received (it didn't challenge anything about the narrative)
|
|
|
+
|
|
|
+### Changing views
|
|
|
+- Sept 2020 clear the narrative is driven by flawed approach of equating case with PCR test and asymptomatic testing
|
|
|
+- Pointed out problem of showing increasing case numbers without understanding changes in tests being performed
|
|
|
+- Diving two numbers is a subversive act of misinformation - called conspiracy theorist
|
|
|
+- Demands to be sacked frm Queen Mary
|
|
|
+- Exposed flaw government data/claims which relied heavily on PCR
|
|
|
+- Concerns about vaccine safety/efficacy
|
|
|
+- Nothing was reviewed / was rejected by pre-print servers
|
|
|
+- Now only on ResearchGate and blogs
|
|
|
+- All videos have been removed (banned from posting videos on YouTube)
|
|
|
+- LEtter into American journal of theraputics - looked at Bayesian model analysis techniques
|
|
|
+
|
|
|
+## Data problems
|
|
|
+### Fundamentals
|
|
|
+#### Definitions
|
|
|
+- Case => Positive PCR test, dies of reasons other than Covid => Covid hospitalization and Covid death
|
|
|
+- All metrics driven by Covid Case
|
|
|
+- - How many asymptomatic?
|
|
|
+- - How many hospitalized for non-COVID?
|
|
|
+- - How many infected at institution?
|
|
|
+- - How many deaths with infection vs from infection?
|
|
|
+
|
|
|
+#### PCR Positive
|
|
|
+Cases include:
|
|
|
+- Virus + symptoms
|
|
|
+- Virus + presymptomatic
|
|
|
+- Virus (never symptomatic)
|
|
|
+- No Virus + symptoms
|
|
|
+- No Virus + no symptoms
|
|
|
+
|
|
|
+#### False-positive rate
|
|
|
+Asymptomatic person tests positive and there's a 1% chance of false positives:
|
|
|
+- 1 / 100 test positive
|
|
|
+Person testing positive likely to have virus?
|
|
|
+- Depends on infection rate
|
|
|
+
|
|
|
+If 1/1000 have the virus
|
|
|
+Test 10,000 people
|
|
|
+- 10 have the virus
|
|
|
+- 100 false positives
|
|
|
+- 10/110 = 9.1% chance positive test is accurate
|
|
|
+Less than 10% chance you really have the virus. The vast majority of cases are false positives.
|
|
|
+
|
|
|
+#### Example
|
|
|
+Cambridge rigorous testing of asymptomatics.
|
|
|
+5000 tested per week for 6 weeks
|
|
|
+43 positives
|
|
|
+7 validated / confirmed cases
|
|
|
+36 false positives
|
|
|
+
|
|
|
+Can't know if those confirmed positive went on to develop symptoms.
|
|
|
+Confirmation test was only PCR, reducing accuracy
|
|
|
+
|
|
|
+#### Lack of confirmatory testing
|
|
|
+- PCR obvious problems
|
|
|
+- What's less well known is that the biggest cause of differences in reporting PCR false positive probabilities is whether the probability assumes a confirmatory test is being carried out:
|
|
|
+- - If the false probability for a single PCR test is 0.5%
|
|
|
+- - All positives require a second test before announcing confirmation
|
|
|
+- - If independent test, probability both are false positive: 0.005 * 0.005 = 0.000025 = 0.0025%
|
|
|
+
|
|
|
+If few people are being tested and you have luxury of confirmatory testing, you will almost see no false positives.
|
|
|
+Confirmatory testing is almost never done
|
|
|
+Most positive PCR test results really are false positives
|
|
|
+
|
|
|
+#### Lighthouse Labs
|
|
|
+- March of 2021 discovery that lighthouse labs were classifying positive cases on the basis of a single target gene
|
|
|
+- WHO and kit manufacturers require minimum of 2
|
|
|
+- In some weeks, 50% of total cases began as false positives (before applying above probability logic again)
|
|
|
+
|
|
|
+#### Impacts
|
|
|
+- 41% of those testing positive never develop symptoms
|
|
|
+- Jan 2020 - Dec 2021:
|
|
|
+- - 4.7% didn't have major underlying comorbidity
|
|
|
+- - Most of these not hospitalized because of COVID
|
|
|
+- Post surgical units showing up to 68% COVID hospitalizations acquired infection post-admission
|
|
|
+
|
|
|
+#### Children (< 18)
|
|
|
+- 5830 hospitalizations (2020)
|
|
|
+- 251 ICU
|
|
|
+- 28 dead
|
|
|
+- 8 confirmed likely caused by COVID
|
|
|
+- All 8 with comorbidities
|
|
|
+- 7/8 had life-limiting condition
|
|
|
+
|
|
|
+#### Comparison
|
|
|
+Daily UK gov cases vs Daily ambulance/emergency calls (NHS dashboard)
|
|
|
+No relationship whatsoever
|
|
|
+
|
|
|
+## Vaccine Efficacy/Safety
|