3. Rising Costs Driven by Aging Population
Prof.Paul Fischbeck,Carnegie Mellon University
4. The Cost of Chronic Disease
Population(2008) Cost(2008)
LTCs
0 1-2 3+
3.6%of 53 M
patients
70%
5.
6. Healthcare Needs BetterTools
o Improve outcomes and reduce costs
o Identify and deliverbest clinical practicefor each patient,every time
and also…
o Shift the balance from treatment to management & prevention
o Engage people as active & informed partners in optimising their health
n = 1
7. The Starting Point
ALL the goods,servicesand content that we consume
will become more personalised over time
Accuratepersonalisationthat is useful to the patient will
driveengagement,trust,behaviourchangeand compliance
10. PersonalisingTherapies & Dose
PERSONALISED MEDICINE 101
More accurate assessment of diagnosis and response to treatment.
Molecular profiling is used to determine the appropriate therapy.
12. RealWorld Personalisation Challenges
o Clinical status
o Phenotype
o Co-morbidities
o Co-prescriptions
o Lifestyle & environment
o (Clonal) heterogeneity
o Polygenic disease
aetiology
PERSONALISED MEDICINE 101
More accurate assessment of diagnosis and response to treatment.
Molecular profiling is used to determine the appropriate therapy.
13. Massive combinatorial analysis of genomic,phenotypic and clinical factors
A collaborationbetweenGenoKeyApSandRowAnalyticsLtd
14. Discover New Disease / Response Associations
Traditional Biomarker Discovery
1 2 3 4 5 6 7 8 9
SNPs
SNPs
Phenotypes
Clinical Data
Co-morbidities
Co-prescriptions
Environment / Lifestyle
Tests / Assays…
Multi-Trait Analysis requires vastly more associations
to be tested between more factors of different types
15. The Jetpack Challenge
Current fastest supercomputer does3 x 1016 operations/sec
At nSNPs= 6 wouldtake 3.1 trillionyears…
At nSNPs= 10 would take3.6 x 1038 seconds= (ageofuniverse)10
(nSNPs)
16. Breast Cancer (BRCA1/BRCA2) Multi-Trait Study
200,000 SNPs for 15,000 people– 4K affected,11K non-affected
Found 5,6 & 7 BRCA1 SNP clusterslinked tobreast cancer risk
1x 7 and 3x 6 SNP clustersassociatedwith diseaserisk (24% ofpop.)
4x 5 SNP clustersassociatedwithprotectiveeffect(36% of pop.)
p-value< 0.001
80 node GPU compute cluster
Initial6 SNP clusterdiscovery– 6 mins
Randompermutations6 SNPs(p-value< 0.001) – 18 hrs
17. Validation Studies
o Identified 4 distinct bipolar disorder subgroups
o 4 distinct 4-SNP clustersamongst bipolar patient population
o Reproducible withWTCCC dataset (479K SNPs,2,000 cases,2,000 controls)
o Clinicallyvalidated into distinctcohorts,e.g.episodicmania,alcohol related
o Identified 100x improvement in PSA test for selected cohorts
o 90% accuracy improved to 99.9% accuracy
o Using combination of SNPs and PSA assaydata
18. o 15,000MND patients / 7,500 controls
o 33% whole genomes sequenced
o 2 petabytes (1,000 2TB hard drives or 3g of DNA!)
o Multi-factor late-onset disease
o Only 5-10% genetically determined heritability
o 6 independent factors required to trigger disease
19. NGS Data Challenge
o Single human genome is ~80GB
o Detection of low penetrance variants
requires at least100x coverage (~1TB)
o Clonal heterogeneity in tumours
o 60M genomes = 60 exabytes = 60,000 PB
o Total stored humandata 2016 = 2,000 exabytes
o 250 mile high stack of
3TB hard drives
20. NGS is not Ready for the Clinic
Low concordance of multiplevariant-calling pipelines:practical
implicationsfor exome and genome sequencing
Genome Medicine (2013), 5 3
Jason O'Rawe12, TaoJiang3, GuangqingSun3, Yiyang Wu12, Wei
Wang4, Jingchu Hu3,Paul Bodily5, Lifeng Tian6, Hakon Hakonarson6,
W Evan Johnson7, Zhi Wei4, Kai Wang89* and Gholson J Lyon129*
22. The Power Shift in Health
o Growing demand for increased
personalisation ofall health services
o Healthcare willbecome
patient centric AND patient mediated
o Health will happen outside healthcare
o Supermarkets, workplace, leisure, gym,
home
o Digital health tools will enablen = 1
23. Personal Responsibility for Health
NCEs & Novel Devices Clinicians& Payers Patients
Therapeutic and lifestyle
programmes
Health Services Health Consumers
Timewithclinician(mins/yr) Timemanagingowncare(mins/yr)
Refractory Cardiometabolic
Patient 80 12,480
CF Patient 240 80,800
Renal Disease Patient ~80 34,320
Self-CareThe New Principal Care,Journal of Ambulatory Care,Nelson etal.
OldValueChain
NewValue
Chain
24. Personalising DiabetesTreatment
o N.I.study on 1,800 insulin injecting adults with type 2 diabetes
o Average userdiagnosed for
10 years,injecting for 5 yrs,
poor A1C compliance (9.4%)
o Reduced drugs load (e.g.DPP-4 inhibitors,SGLT-2 inhibitors,GLP-1 agonists)
o 96% patient satisfaction,90% patient retention,28% cost of care savings
http://hygieia.com/case-studies/
25. Smoking Cessation
o 38 NZ smokers given genetic
test assessingpersonal risk
for developinglung cancer
o Average:
o age 56
o 35 yrs smoking
o 17 cigarettes/day
o 63% M :37% F
Respiragene (NZ)study funded by support from Synergenz BioScience Ltd (www.synergenz.com)
29. HealthySwaps (www.healthyswaps.diet)
o Truly personalised advice on drug,disease & food interactions
o Real-time,all on user’smobile phone/tablet
o No IBMWatson ;-) / server
o Accounting for all potential interactions and health goals for each user
o All personal information is fully encrypted and only ever stored on a
patient’s mobile device (phone/tablet)
o User’s personal information isnever seen by anyone
30. BetterTools for Healthcare
o Clinical and patient decision support tools
o At the point of care /
in day-to-day life
o Using full power of complex,
multi-trait knowledge models
o Improving patient outcomes
o Lowering the cost of care
provision