Post on 07-Jul-2020
26/05/2020
1
The odd one out?
John Arbuthnot Meteorological models
Claude Shannon
provided a formal
definition of
information based on
probability and a
receiver – allowing
information to become
a scientific quantity –
and laid the
foundations of the
information age
“400,000 fewer children in poverty than in 2010”
Boris
26/05/2020
2
“400,000 fewer children in poverty than in 2010”
The data
But child poverty increases with benefit freezes and the
loss of local government services!
The number of children living in poverty is estimated to
have increased by 500,000 between 2010/11 and
2017/18 under the DWP’s relative low income measure
The Institute of Fiscal Studies (IFS), the Joseph Rowntree
Foundation and the Social Metrics Commission all agree
that child poverty has worsened
IFS predict an increase in child poverty in future years
Regardless, 3.5 Million children living in absolute
poverty does not exactly constitute ‘success’
Reality check
Inequalities CMO report
“we are not doing as well as
we should”
3-4% have developmental
disorder (risk factor for poor
educational attainment)
health barriers to education
short-term costs estimated as
£1.58 billion per annum
annual long-term costs
estimated at £2.35 billion
Annual report of the
Chief Medical Officer 2012
Our Children Deserve Better:
Prevention Pays
What the data show The problem
The system is broken
It’s complex and dynamic...
But our actions – e.g.
government policies - are
often not tailored for
complex dynamical systems
26/05/2020
3
Act early
Schools
Health services
Digital Triangle
Data Science
Artificial Intelligence Immersive Technologies
Innovative history
Baseline
• Lifestyle
• Environment
• Medical (physical & mental health)
• Socio-economic
• Genetic
• Biological
Routinely collected
data
• Health visitor/school nurse
• Hospital visits
• GP records
• Bradford Education
• Social care
Full Sweep @
4-5yrs
8-10yrs
SECONDARY
• Education
• Environmental
• Medical
• Cognitive
Lifestyles,
obesity &
activity
Asthma,
allergies &
infection
Pollution &
green
space
Learning &
behaviour
Born in Bradford
https://borninbradford.nhs.uk/research/documents-data/
Connected Bradford & Airedale Database – De-identification Process 1v1
Ap
ollo
cBra
dfo
rd D
atab
ase
Man
ager
cBra
dfo
rd D
ata
An
alys
tN
ECS
DfE
/LA
Dat
a P
rovi
der
s P
seu
do
at
sou
rce
BTH
FT a
nal
yst
Bradford District and Craven
18.Build the data extract from
Production Server
16.Extract Transform Load
Server
19.Researcher Specific Data
1.Salt Key is sent to named Caldicott Guardians or
Designated Officers upon written request
10.Designated Officer(s) receives
Salt key
13.De-identify
Primary Care Data
15.cBradford ETL Server
17.Production Server
12.Pseudonymise NHS Number in
Primary Care Data
14.Pseudonymised & De-identified
Primary Care Data sent via SFTP
11.Create EDRN
4.De-identify NCMP /
Adult Social Care
2.Caldicott Guardian receives Salt key
3.Pseudonymise NHS Number in NCMP / Adult
Social Care Data
5.Pseudonymised & De-identified
NCMP / Adult Socail Care Data sent via
Kiteworks
20.Send identifiers & EDRN to DfE
22.DfE receives identifiers EDRN
23.Match Education records using
identifiers
24.Create file of EDRN & Matched
Education Records
25.Send EDRN & Education Records
to BTHFT
27.Match Education & Primary Care data
using EDRN
26.Receive Pseudo ID & EDRN from
Apollo
31.Researcher Specific Data
30.HERD Research Database
21.Send Pseudo ID & EDRN to
cBradford Database Manager
8.De-identify
BTHFT Data
6.Caldicott Guardian receives Salt key
7.Pseudonymise BTHFT Data on BTHFT Source
Server
9.Pseudonymised & De-identified BTHFT
Data copied to Secure Drive
28.HERD ETL Server
29.Extract Transform Load
Server
Approvals:Health Research Authority (HRA) Confidentiality Advisory Group (CAG) REC & IRAS Yorkshire & Humber Bradford Leeds Research Ethics CommitteeWhite Rose Universities Ethics Committees
TPP
LA
BiB data Double disadvantage
OCT
31-32 33-34 35-36 37-38 >40<31
NOV MAY JUN JUL AUGDEC JAN FEB MAR APR
Months of birth within the academic year
Gestational age (weeks)
Odds relative to September birth
Odds relative to birth in 39-40th week
Odds
ratio
(log)
Odds
ratio
(log)2
3
4
1
2
3
8
1
26/05/2020
4
Visual screening
1.0
0.9
0.8
4%
0.092%
0.3
0.7
2%
0.2 2%
Missing in action
866 children (16%) referred for further investigation
534 referred to optometrist
332 referred to Hospital Eye Service
479 (55.3%) attended the first appointment
371 (42.84%) failed to attend
345 were sent a recall appointment
111 (32.17%) attended the recall appointment
[HES coverage improved by10.54% (60.84% to 71.39%) and
Optometrist coverage improved by 14.23% (51.87% to 66.10%)]
234 lost in
action
Seeing and vocabulary
Letter identification affected by poor eyesight
Score 2% every one line logMAR reduction: CI -2.17 to -1.13
Improving autism care
We used BiB data to show that the educational assessments
conducted on all children in their first year of schooling (the
Early Years Foundation Stage Profile) can identify children
at risk of autism
examined EYFSP scores for13,500 children within BiB
educational data linked to routine health in 8,854 children
Creative interventions
Early detection and support for autism
Glasses in classes - world’s largest RCT
Data driven service organisation World’s largest handwriting RCT
https://borninbradford.nhs.uk/research/documents-data/
Connected Bradford & Airedale Database – De-identification Process 1v1
Ap
ollo
cBra
dfo
rd D
atab
ase
Man
ager
cBra
dfo
rd D
ata
An
alys
tN
ECS
DfE
/LA
Dat
a P
rovi
der
s P
seu
do
at
sou
rce
BTH
FT a
nal
yst
Bradford District and Craven
18.Build the data extract from
Production Server
16.Extract Transform Load
Server
19.Researcher Specific Data
1.Salt Key is sent to named Caldicott Guardians or
Designated Officers upon written request
10.Designated Officer(s) receives
Salt key
13.De-identify
Primary Care Data
15.cBradford ETL Server
17.Production Server
12.Pseudonymise NHS Number in
Primary Care Data
14.Pseudonymised & De-identified
Primary Care Data sent via SFTP
11.Create EDRN
4.De-identify NCMP /
Adult Social Care
2.Caldicott Guardian receives Salt key
3.Pseudonymise NHS Number in NCMP / Adult
Social Care Data
5.Pseudonymised & De-identified
NCMP / Adult Socail Care Data sent via
Kiteworks
20.Send identifiers & EDRN to DfE
22.DfE receives identifiers EDRN
23.Match Education records using
identifiers
24.Create file of EDRN & Matched
Education Records
25.Send EDRN & Education Records
to BTHFT
27.Match Education & Primary Care data
using EDRN
26.Receive Pseudo ID & EDRN from
Apollo
31.Researcher Specific Data
30.HERD Research Database
21.Send Pseudo ID & EDRN to
cBradford Database Manager
8.De-identify
BTHFT Data
6.Caldicott Guardian receives Salt key
7.Pseudonymise BTHFT Data on BTHFT Source
Server
9.Pseudonymised & De-identified BTHFT
Data copied to Secure Drive
28.HERD ETL Server
29.Extract Transform Load
Server
Approvals:Health Research Authority (HRA) Confidentiality Advisory Group (CAG) REC & IRAS Yorkshire & Humber Bradford Leeds Research Ethics CommitteeWhite Rose Universities Ethics Committees
Connected Bradford
26/05/2020
5
School admissions Place based working
Not the same place!
White British
heritage -
wealthy
South Asian
heritage -
deprivation
White British
heritage -
deprivation
January 10th 2020
Working with our partners to use data to support
place based working across the Region
11,130 people live in Holme Wood
(28.3% under16 years of age)
84.3% White British plus Gypsy
Highest % of NEET in Leeds City Region
22% long-term illness and on benefits
Life expectancy <10 years below UK
Mental health a major problem
High rates of crime