dacadoo is proud to be referenced by Swiss Re - Center of Global Dialogue in report 'Healthcare revolution: Big data and smart analytics'.
dacadoo is referenced in the section "Sensor innovations driving the digital health revolution"
3. Swiss Reā Conference report: Healthcare revolution: Big data and smart analyticsā 1
Table of contents
Introduction 3
From diagnosis to personalised prognosis:
Will better information lead to better decisions? 4
Sensor innovations driving the digital health revolution 6
Reducing the burden of chronic disease through
remote monitoring and management 8
Using common data to make uncommon predictions 10
The wireless future of medicine:
How digital revolution will create better healthcare 12
Healthcare analytics:
From reactive to proactive decision-making 14
Designing digital interventions using behavioural economics 15
Using predictive analytics to create pre-approved life insurance 16
Panel discussion on healthcare revolution:
What is next? 17
Organisers 19
4. 2ā Swiss Reā Conference report: Healthcare revolution: Big data and smart analytics
6. 4ā Swiss Reā Conference report: Healthcare revolution: Big data and smart analytics
From diagnosis to personalised prognosis:
Will better information lead to better decisions?
Karin Frick, Head of Think Tank, and Member of the Executive Board,
GDI Gottlieb Duttweiler Institute
Prediction was once the realm of priests and magicians. In todayās society it is
increasingly a tool of planning. Where once we had post-illness diagnosis of a
condition, the future promises pre-illness prognosis.
Our increasing faith in prediction is based on technology. A body of data-driven tools
is capable of discovering and analysing patterns so that past correlations can be
used to forecast likely future outcomes. Predictive technologies, which include data
mining, neural networks, and system modelling and simulation, have been applied
to the study of weather systems, traffic patterns, stock markets, epidemiology,
consumer behaviour, terrorist activity, and many other areas of study where there
can be a significant number of variables.
Technology and our ability to record data is expanding exponentially; and healthcare
is seen as one of the primary beneficiaries. Entrepreneurs are creating apps that
can run on smart phones claiming to be capable of predicting conditions from
depression to sport injury. These metrics not only benefit the individual, but also the
wider healthcare industry. With the data gleaned from these devices, insurance
companies, employers and healthcare providers can view a comprehensive picture
of an individual and a populationās health ā one that is more accurate and
trustworthy than a first-person narrative.
This increase in predictive ability should be a force for good; but human reactions
may be unpredictable. Prognosis could provoke change; equally it could provoke
over- or underestimation of the risk, or ā if the prognosis is not positive ā to ignore
it altogether. We already know that humans have a tendency to drift towards
hyperbolic discounting. The risk of a terrorist attack, which statistics tell us is highly
unlikely, is seen by most respondents as being far higher than diabetes ā which
statistics suggest is a much likelier fate. Moreover, most people have an inability to
imagine how they will age and how their preferences and personalities will change.
The most effective way for humans to take on and realise information is through
feedback loops. The first stage is evidence, the actual data; then comes the
prognosis based on the data; the relevance of the prognosis has to be realized in
its social and physical context; the consequences of the prognosis are understood;
and the individual finally acts according to the previous four steps. The circle is
started again.
Prognosis
Evidence Relevance
Action Consequences
Source: GDI Gottlieb Duttweiler Institute
Figure 1:
The power of feedback
7. Swiss Reā Conference report: Healthcare revolution: Big data and smart analyticsā 5
These stages can be identified within the use of fitness armbands that record
physical movement. Evidence (steps) leads to prognosis (expected gains in fitness);
leads to relevance (frequently in a gaming context, which encourages participation);
leads to consequences (feel better, weight loss); and finally action (take more steps
in a day).
This does not mean there are not questions about predictive technology. Human
happiness is not a universal defined quantity. Some may feel uncomfortable in the
knowledge that our future is already defined. Moreover, it is difficult to know what to
do with prognoses that may not be beneficial; such as a predisposition to criminal
behaviour or a degenerative condition. Predictions can be self-fulfilling. As W.I.
Thomas and D.S. Thomas suggested in 1928, āIf men define their situations as real,
they are real in their consequence.ā
8. 6ā Swiss Reā Conference report: Healthcare revolution: Big data and smart analytics
Sensor innovations driving the digital health revolution
Peter Ohnemus, President and CEO, dacadoo
The quantified self ā the ability to record how we move, eat and sleep ā is already
upon us. It stands at the convergence of a number of societal trends. The first is that
of the smart device. Smartphones now have the computing power of 1980s main
frame computers. They are ubiquitous and the industry is keen to push sensor
technology which records our daily activity. There is an explosion in the production
of sensory devices currently. It comes at a time when social networking is increasingly
forming a platform of interaction; and where the phenomenon of gaming, either
against the computer or against social networks, is growing.
Source: iStock
Perhaps of greatest relevance is that mobile health is gaining traction as overall
healthcare costs spiral; the world spends about USD 6 trillion on healthcare, a factor
of four compared to what it spends on food. Healthcare as a percentage of GDP is
rising towards 20% in the US; with predictions that, with an elderly society and more
chronic disease, it could rise to 30% in coming decades. These healthcare costs are
currently a liability for insurance companies and the government; increasingly there
will be a trend to transfer healthcare costs towards employers. Studies suggest that
using mobile health to contribute to āright livingā could save US healthcare costs of
USD 70ā100 billion per year.
In order for mobile health to be effective, it has to be easy to access ā via a mobile
device ā it has to be fun and it has to be relevant. Relevance can be achieved by
making data easy to understand; and platforms such as dacadoo can formulate a
single health score from an array of data points. Some consumers will have concerns
about data privacy; however, data privacy is as much an issue with analogue records
as it is with digital. Many mobile health providers take the question of data extremely
seriously.
Figure 2:
Smart devices can measure heart rate,
blood oxygen levels, temperature and
movement recognition. They will only
grow in sophistication.
9. Swiss Reā Conference report: Healthcare revolution: Big data and smart analyticsā 7
Source: dacadoo
Mobile health will be a massive growth industry; according to a report from BCC
Research it will be worth USD 50 billion by 2020 and the sector is gaining from
large inflows of venture capital. The Obamacare programme will influence this
development further. As of 2014, employers may use up to 30% of health insurance
premiums to provide outcome based wellness initiatives. While it is still likely that the
insurance model will stay social in its broad construction, health indicators will
increasingly drive individual premiums, most probably initially in the form of rebates
for verifiably healthier customers.
Figure 3:
The dacadoo health score ā a number from
1 (low) to 1000 (high) represents your
current health and fitness status in real-time.
When tracked over time, the dacadoo health
score offers a good directional relative
indicator of how your health and fitness is
improving or deteriorating.
10. 8ā Swiss Reā Conference report: Healthcare revolution: Big data and smart analytics
Reducing the burden of chronic disease through
remote monitoring and management
Laurence Jacobs, Senior Research Scientist, University of Zurich Medical School
The traditional approach to the management of chronic diseases is not optimal from
a medical perspective, and it is extremely expensive. Moreover, the worldwide
growth in cases of chronic disease continues to increase at a very fast pace.
Reasonable cost estimates place the total financial burden caused by chronic
disease in the several hundreds of billions of dollars annually. This situation is
untenable in the long run. Left unattended, this problem is such that in the not too
distant future, no society will be able to afford the cost of caring for its ailing
population. The traditional approach to this problem simply does not scale well.
Fortunately, there are alternatives to the traditional approach. These alternatives, at
present mostly in the development and testing phases, are not only much cheaper,
but they have the potential of being also better for the patient from a medical
perspective.
The current opportunity was born not only of necessity, though that has played an
important role, but also from the confluence of the general populationās interest in
health. Companies have developed small, accurate and inexpensive biosensors.
These have led to a growing availability of good quality data that can be used to
derive accurate models that can generate alerts, or even trigger devices to react to
critical changes in the one or more parameters being monitored.
Diabetes is a prominent example. As far as growth, it is estimated that there will be
around 250 million sufferers worldwide by 2030, more than double the amount
estimated in 2005. A key component of the process of managing diabetes is to
measure the level of glucose in the blood several times a day. With the technology
of a few years ago, this process is painful, expensive and cumbersome, requiring the
extraction of blood and the use of portable meters. However, current technology
already allows for a reasonably practical way to measure glucose continuously using
a sensor that is implanted subcutaneously. Even better, several start-ups are
announcing systems to measure glucose continuously without the need to extract
any blood at all. These sensors, several using light, or estimating the levels of blood
glucose by analysing tears or saliva, will soon become commercially viable. These
systems will not only be simpler and cheaper, but they will also lead to better
methods of treatment.
There are currently many clinical trials underway that aim to test integrated
platforms, running on smartphones, that measure, analyse, and report on multiple
continuous measurements of a potentially large number of important biometrics that
promise to optimise the treatment of several chronic diseases. Patients and their
doctors can be informed in real time on effective treatment change, and alert on
critical risk factors. This would have been impossible only a few years ago, and it
will eventually revolutionise the management of chronic disease.
11.
12. 10ā Swiss Reā Conference report: Healthcare revolution: Big data and smart analytics
Using common data to make uncommon predictions
Ben Reis, Director of the Predictive Medicine Group at Harvard Medical School and
the Childrenās Hospital Informatics Program
We are living in the age of many unknowns. In two major respects, we are entering
new territory for human evolution. One of those is age. Evolution has focused on
keeping us healthy through reproduction and child-rearing years, but many people
now live well beyond this stage. Those extra years bring co-morbidities and chronic
conditions. Another novelty is urban living. There have been town dwellers for
thousands of years; but never have we had more city dwellers than rural farmers.
That implies different environments and different patterns of physical activity. The
more we know about the effects of these new trends in human existence, the more
we can seek to positively influence our health.
Previous healthcare revolutions have not proved to be quite as revolutionary as
initially hoped. Although decoding the genome was a ground-breaking achievement,
rather than providing a clear blueprint of our genetic selves, it revealed how much
we did not know about the interaction of genes in determining human traits and the
existence of epigenetics, the interaction of genes with their environment. Similarly,
whether the potential treasure trove of data that accompanies the smartphone
revolution will enhance our capacity to predict health status remains to be seen.
We already have, however, rich sources of data available to us. One source is claims
submitted from providers to insurers to cover medical costs. These data provide
information on the frequency of individual visits to the doctor, the conditions the
individual suffers from, the geographic placement of claimants, the prescriptions or
treatments undertaken by the doctors, and the effectiveness of these treatments as
registered by the need of the individual to return to the doctor subsequently. On the
whole, insurance claims have produced some remarkable insights:
Ģ¤Ģ¤ Behavioural models: Health records provided a surprisingly accurate prediction as
to those individuals who might be more susceptible to being victims of domestic
violence.
Ģ¤Ģ¤ Epidemiological models: Insurance claims have proved a good proxy in the
identification of disease clusters and outbreaks.
Ģ¤Ģ¤ Predictive drug affects: The safety of drugs is generally measured in comparison
to a reference drug. Using network models, adverse drug effects can be detected
and even predicted years in advance.
13. Swiss Reā Conference report: Healthcare revolution: Big data and smart analyticsā 11
More predictive Less predictive
Alcohol- and substance-
related mental disorders
Anxiety-
Somatoform-
Dissociative-
and
Personality-
Disorders
Superļ¬cial injury;
contusion
Sprains and
strains
Residual codes;
unclassiļ¬ed
Other
injuries
from
external
causes
Poisoning Open
wounds
Burns
Other mental
conditions
Aļ¬ective
disorders
History/
screening
mental
disorder
Headache,
including
migraine
Epilepsy,
con-
vulsions
Asthma Diseases of
female genital
organs
Schizo-
phrenia
and related
disorders
Factors inļ¬uencing
health care
Back problems
Tooth,
jaw
disorders
Liver
disease
Other
psychoses
Viral infection
Source: Reis B Y et al. BMJ 2009; 339: bmj.b3677
There are other rich sources of existing public data. One of these is search engine
query data. One study suggests an inverse relationship between the availability of
abortion services in a particular US state, and Google searches for abortion services.
This may suggest that demand for abortion is relatively constant across different
states, it is the supply side that accounts for different abortion search rates.
Figure 4:
Some risk factors for domestic violence in
women, as indicated by insurance claims.
This is one of a number of innovative means
of capturing and visualising data enabled by
modern data analysis software.
15. Swiss Reā Conference report: Healthcare revolution: Big data and smart analyticsā 13
Source: www.ga-project.eu
Such sensors are already being built into wearable technology, such as armbands,
smart garments and smart patches. They will be able to be transferred to lenses for
the eyes and onto the skin in the form of skin tattoos. They should aid users in
ultimately sensing and diagnosing conditions; allowing autonomy of decisions;
security and privacy of the personal data and providing the platform for better
decisions. Multiple sensor technologies have the ability to provide a more holistic
health picture; which can be complemented by sensors monitoring physical
manifestations of the individualās emotional state. The final goal is a paradigm shift
from prescription to prevention, enabling new solutions for more sustainable
healthcare models.
Figure 5:
The Guardian Angels European-wide
partnership is teaming up to develop ever
more compact and streamlined technology
that slips seamlessly into our daily lives.
17. Swiss Reā Conference report: Healthcare revolution: Big data and smart analyticsā 15
Designing digital interventions using behavioural economics
Dominic King, Clinical Lecturer in Surgery and Behavioural Economist,
Imperial College London
Behavioural economics is attracting policy makersā attention in countries including
the UK, France, Australia and the USA. By incorporating insights from psychology
with the laws of economics, behavioural economists have demonstrated how peopleās
behaviour can be strongly influenced by small changes in the context or environment
in which choices are made. The basic insight of behavioural economics is that
human behaviour is not guided by logic of a supercomputer, but is determined
rather by our fallible and very human brains. Psychologists and neuroscientists have
recently converged on a dual process model of decision-making. On one side the
automatic system (System 1) provides fast, unconscious, intuitive, decision-making
often based on mood or emotion. On the other, the reflective system (System 2) is
slower, conscious, reflective, and rational. Behavioural economists are now consulted
at the highest levels of government and industry, in an effort to target automatic
decision-making and move away from targeting rational processes through
information and price signals. Approaches targeting automatic processes are
popularly called nudges after the influential book Nudge by Richard Thaler and
Cass Sunstein.
The mnemonic MINDSPACE (messenger, incentives, norms, defaults, salience,
priming, affect, commitment, and ego) seeks to capture effects acting largely but not
exclusively on automatic processes. Norms, for example, suggests we are strongly
influenced by what others do; while default suggests individuals go with pre-set
options (prompting the use of āopt outā rather than āopt inā defaults for choices such
as organ donor cards). MINDSPACE was developed by the UK Cabinet Office and is
widely used across the private and public sector in the UK.
Cue Behaviour
Messenger We are heavily influenced by who communicates information to us.
Incentives Our responses to incentives are shaped by predictable mental shortcuts
such as strongly avoiding losses.
Norms We are strongly influenced by what others do.
Defaults We āgo with the flowā of pre-set options.
Salience Our attention is drawn to what is novel and seems relevant to us.
Priming Our acts are often influenced by sub-conscious cues.
Affect Our emotional associations can powerfully shape our actions.
Commitments We seek to be consistent with our public promises, and reciprocate acts.
Ego We act in ways that make us feel better about ourselves.
Source: Institute for Government and UK Cabinet Office
There is huge interest in applying insights from behavioural economics to influencing
health-related behaviours. This is because the consequences of suboptimal decision-
making is so substantial both in terms of morbidity and mortality and also financial
costs. Delivering behaviour change interventions on a population scale may be aided
by the increasing ubiquity of smartphones and tablet devices. The majority of people
in developed countries (and increasingly low- and middle-income countries) now
have access to these devices and they are usually with people at work and at home.
These devices can be used to both monitor health and deliver healthy nudges.
Interventions may be as simple as a text message or something much more complex.
Smartphone apps are now being widely used to improve outcomes in areas
including medication adherence, diabetes and smoking cessation.
At Imperial College London we have shown how MINDSPACE interventions
delivered over mobile phones can help adolescents maintain weight loss after
attending residential weight loss camps. We have also shown that a simple change
in the text of an SMS message can significantly decrease the number of people
who miss appointments in public hospitals in London. There is an opportunity to do
much more.
Figure 7:
MINDSPACE
The nine most robust effects that operate
largely, but not exclusively, on the automatic
system affecting human behaviour.
18. 16ā Swiss Reā Conference report: Healthcare revolution: Big data and smart analytics
Using predictive analytics to create pre-approved life insurance
Edward Leigh, Innovation Lab Lead, Protection, Aviva LifeāāPensions UK Limited
William Trump, Predictive Underwriting Consultant, LH Products, Swiss Re
There have been fears that the use of predictive analytics, combined with the
resources now provided by big data, could be used by insurers to restrict or select
their risk pool. That has not been the case; but insurers have been using predictive
analytics on the marketing side, most notably with an application to life insurance.
Life insurance is an opaque product that frequently puts consumers off. They do not
want to think of the consequences ā death ā and do not want to be faced with the
hassle of form filling and underwriting. Yet most life insurance applicants are healthy.
Predictive analytics is being used to pre-approve life insurance products for certain
individuals and provide them with greater clarity and ease of process in acquiring life
insurance.
āLife insurance is
complicatedā
āItāll take ages to
applyā
āLife insurance
isnāt for meā
āItāll be really
expensiveā
āI donāt like
declining
customersā
āThere are lots of
complicated
wordsā
These insights can be used to
reduce the amount of traditional underwriting
(where there is data)
Predictive underwriting: the use of non-medical
data held on customers to reach a view
on their health
And we are in a data rich environment
Similar challenge faced by banks with loans
and credit cards
But most customers are in good health
Current sales process for life insurance is a
barrier for customers and staff
Easy and
straightforward
Quicker and
slicker
Personalised
communications
Actual price, not
indicative price
Pre-approved
customers to
boost confidence
No medical
terminology
Source: Leigh, E., Trump, W. (2014)
The data being used in the predictive modelling comes from Avivaās banking
partners. The data includes over 150 variables attached to bank accounts; for
example: age, occupation, house owner, direct debits, credit card payments and
bank ATM withdrawals. Together, these predictors form the algorithm that can
score any banking customer in terms of ālikelihood to be given standard rates at
underwritingā ā that is, that they are healthy.
The result is a one question questionnaire (whether the respondent has cancer or
diabetes). It helps front line colleagues, with consultations falling from as long as
90 to as quick as 15 minutes, as well as being simpler. It helps customers engage
with life insurance quickly and easily. The success of life products could see similar
pre-approved techniques being rolled over to lines such as critical illness.
Figure 8:
A case study for pre-approved life
insurance: the problem (red),
the logic (grey), the aim (green)
20. 18ā Swiss Reā Conference report: Healthcare revolution: Big data and smart analytics
Panel discussion and concluding remarks
Concerns for data security could lead to the danger of data becoming placed in silos
and not shared to provide a more holistic picture of an individualās or a populationās
health. One possible means to better sharing would be to entrust an individualās data
in their own databanks ā allowing the individual to share their data as they wish.
However, it would be equally unwise to suggest that data offers a healthcare
panacea. Risk will still be entailed. Equally, it would be wrong to think that medical
treatments can be reduced to a set of standardised steps. Every patient is different;
and there is a little art in the way a doctor prescribes treatments, not just science.
Data is a powerful tool, but needs to be used alongside a human touch.
The day began by suggesting that data would allow prognosis rather than diagnosis
in the field of health. The conclusion has to be that the potential is there. New
technologies exist in abundance; while new data modelling techniques can take
advantage of existing data sources. However, concerns over the appropriate use of
data and the appropriate approvals for the use of data clearly exist. Trust will be key
to both of these factors. The balance must rest with the consumer: they must feel the
benefit; equally they increasingly may feel responsibility with regards to data sharing.
21. Swiss Reā Conference report: Healthcare revolution: Big data and smart analyticsā 19
Organisers
The Swiss Re Group is a leading wholesale provider of reinsurance, insurance and
other insurance-based forms of risk transfer. Dealing direct and working through
brokers, its global client base consists of insurance companies, mid-to-large-sized
corporations and public sector clients. From standard products to tailor-made
coverage across all lines of business, Swiss Re deploys its capital strength, expertise
and innovation power to enable the risk taking upon which enterprise and progress
in society depend.
The GDI Gottlieb Duttweiler Institute is a non-profit organisation conducting
scientific research in the social and economic fields. The institute studies mega-
trends and countertrends and draws up scenarios for the future. The GDI is also a
meeting place: it hosts leading thinkers and decision-makers at its conferences, and
it makes its spaces and infrastructure available for corporate or private events.
The Swiss Re Centre for Global Dialogue is a platform for the exploration of key
global issues and trends from a risk transfer and financial services perspective.
Founded by Swiss Re, one of the worldās largest and most diversified reinsurers, in
2000, this state-of-the-art conference facility positions Swiss Re as a global leader at
the forefront of industry thinking, innovation and worldwide risk research. The Centre
facilitates dialogue between Swiss Re, its clients and others from the areas of
business, science, academia, and politics.