Big data in medical science and healthcare management

Thilo Weichert12 Medical Big Data and Data Protection  139 Sebastian Krolop and Henri Souchon 13 Big Data in Healthcare from a Business Consulting Accenture Point of View  151 16 Quant

Peter Langkafel (Ed.)

Big Data in Medical Science and Healthcare Management

Advanced Data Management

Big Data in Medical

Science and Healthcare Management

Diagnosis, Therapy, Side Effects

Published by

Dr med Peter Langkafel, MBA



Set-ISBN 978-3-11-044575-6

Library of Congress Cataloging-in-Publication Data

A CIP catalogue record for this book has been applied for at the Library of Congress.

Bibliographic information published by the Deutsche Nationalbibliothek

The Deutsche Nationalbibliothek lists this publication in the Deutsche Nationalbibliografie; detailed bibliographic data are available in the Internet at http://dnb.dnb.de.

With kind permission of medhochzwei Verlag.

The german edition of Langkafel, Peter (Ed.), “Big Data in Medizin und Gesundheitswirtschaft – Diagnosen, Therapien, Nebenwirkungen ” has been published by medhochzwei Verlag, Alte Eppelheimer Straße 42/1,

69115 Heidelberg, Germany

© 2014 medhochzwei Verlag GmbH, Heidelberg

www.medhochzwei-verlag.de

Translation: Thorsten D Lonishen, CL-Communication GmbH, Germany: www.cl-communication.com

© 2016 Walter de Gruyter GmbH, Berlin/Boston

Cover image: Shironosov/iStock/Thinkstock

Typesetting: Lumina Datamatics

Printing and binding: CPI books GmbH, Leck

♾ Printed on acid free paper

Printed in Germany

www.degruyter.com

For Gudrun, Arved, Napurga and Anio.You are my best Big Datas!

Autopilot and “Doctor Algorithm”?

The first flight with the help of an autopilot was shown at the world exhibition inParis in the year 1914: Both the vertical as well as the horizontal directions of flightwere controlled by two gyroscopes These were driven by a wind-powered generatorlocated behind the propeller What was a fascinating novelty at the time is todaytaken for granted Nowadays, airline pilots on an average flight actually pilot theplane by themselves for only about 3 minutes.¹ The rest is controlled by modernautomation using a variety of sensors and onboard computers

Most readers will probably not pilot modern aircrafts very often– but autopilots

in modern cars have already become a part of daily life, and these days most driverswill no longer want to do without them Initially, digital maps may have led to somebizarre accidents– such as driving into a lake on a clear day But in modern passen-ger cars Big Data is taking on an ever increasing part of driving behavior: speed, dis-tance, braking behavior, directional stability, etc The range of auxiliary functions isconstantly growing, which is also true for their acceptance A modern car todayincorporates more information technology than the Apollo rocket, which enabled amanned mission to the moon in 1969

Medicine has evolved over the last 100 years – but medical autopilots are notavailable, or should we say not yet available? For some, this may be an inspiringvision of the future – for others the damnable end of treatment with a “humantouch.” Particularly in the field of medicine, a huge, unmanageable amount of infor-mation is generated every day– but computers as coaches and not as data servants:

Is that vision a long way off or have we almost reached it already?

The issue concerns much more than mere technical feasibility If we want tounderstand Big Data in medicine, we have to take a broader view of the matter Inthis book, more than 20 different experts, from the broader field of and around medi-cine, have written articles concerning completely different aspects of the issue– thusincluding the conditions and possibilities of“automation of everyday (medical) life.”Roughly 100 years ago the philosopher and mathematician Alfred North White-head described civilization as something“that develops by increasing the number ofimportant tasks that we can perform without thinking about them.”²

However, the impact in the 21st century is quite different:“Automation does notsimply replace human activities but actually changes them, and often does so in amanner that was neither intended nor envisaged by its developers.”

1 Nicholas Carr: Die Herrschaft der Maschinen: Blätter für deutsche und internationale Politik,

2/2014.

2 Raja Parasuraman, quoted from: Carr: Die Herrschaft der Maschinen: Blätter für deutsche und

inter-nationale Politik, 2/2014.

Big Data– in medicine, too – may be changing the world more than we can yetunderstand or want or wish to acknowledge.

Special thanks goes out to all authors for their contributions to this scope: The texts are hopefully also reflected in each other We often see kaleido-scopes as a“mere” children’s toy With a little technical skill, however, they canalso be used as a“microscope” or “telescope” – but even here, software has alreadybeen developed that can simulate these effects…

kaleido-I hope you enjoy the reading experience and may your insights be small or Big!

Peter Langkafel

Peter Langkafel

1 Intro Big Data for Healthcare?  1

Schepers Josef and Martin Peuker

2 Information Management for Systems Medicine – on the Next

Digital Threshold  33

Albrecht von Müller

3 Some Philosophical Thoughts on Big Data  45

Thomas Brunner

4 Big Data from a Health Insurance Company ’s Point of View  53

Harald Kamps

5 Big Data and the Family Doctor  63

Alexander Pimperl, Birger Dittmann, Alexander Fischer, Timo Schulte, Pascal Wendel,Martin Wetzel and Helmut Hildebrandt

6 How Value is Created from Data: Experiences from the Integrated Health Care System, “Gesundes Kinzigtal” (Healthy Kinzigtal)  69

Rainer Röhrig and Markus A Weigand

Karola Pötter-Kirchner, Renate Höchstetter and Thilo Grüning

8 The New Data-Supported Quality Assurance of the Federal Joint Committee: Opportunities and Challenges  101

Werner Eberhardt

9 Big Data in Healthcare: Fields of Application and Benefits of SAP

Technologies  115

Axel Wehmeier and Timo Baumann

10 Big Data – More Risks than Benefits for Healthcare?  123

Marcus Zimmermann-Rittereiser and Hartmut Schaper

11 Big Data – An Efficiency Boost in the Healthcare Sector  131

Thilo Weichert

12 Medical Big Data and Data Protection  139

Sebastian Krolop and Henri Souchon

13 Big Data in Healthcare from a Business Consulting (Accenture) Point of View  151

16 Quantified Self, Wearable Technologies and Personal Data  197

Axel Mühlbacher and Anika Kaczynski

17 “For the Benefit of the Patient” … What Does the Patient Say

to That?  213

Peter Langkafel

18 Visualization – What Does Big Data Actually Look Like?  223

Peter Langkafel

19 The Digital Patient?  231

Publisher and Index of Authors  235

Glossary  241

Testimonials  247

Peter Langkafel

1 Intro Big Data for Healthcare?

Content

1 Big Data  2

1.1 … or “Little Sexy Numbers”?2

1.2 The First Mister Big Data?3

1.3 Mr Big Data in Medicine?4

1.4 Mr Interpret Big Data6

1.5 The Only Statistics You Can Trust, Are…7

1.6 Mr Understand Data8

1.7 Datability8

1.8 Big Data in the Bundestag9

1.9 The Three Vs of Big Data9

1.10 Can Medicine Learn from Other Industries?10

2 Big Data Framework in Medicine  11

2.1 Fields of Action in Medicine13

2.2 It’s Time for Big Data15

2.3 The Example of Google Flu Trends17

2.4 Big Data Adoption Model in Medicine18

3 Big Data in the Context of Health – More Than Old Wine in New Bottles?  19

3.1 The Difference to Classic Clinical Analysis (Old Wine in New

Bottles?)20

3.2 Is Big Data in the Hands of Medical Controlling Degenerating to a DRG

Optimization Technology?20

3.3 Are Hospitals Prepared for Big Data?21

3.4 Are Internal Hospital Big Data Analyses Data Protection Cliffs, and if so,

Where Exactly?21

4 Personalized Medicine and Big Data  22

4.1 The (Genetic) Human in Numbers23

1 Big Data

Can you remember the first time you heard the term Big Data? This buzzword hasrecently found its way into event titles, panel discussions, articles and backgrounddiscussions at a phenomenal rate How is it possible that two little words could andstill can celebrate such global success?

New terms always act as projection screens onto which the many different gonists and players can interpret their version and vision of things The wider and moreemotional the interpretation of such neologisms, the more promising the word creation

prota-is in itself Big Data means fear, fascination and overestimation at the same time.Perhaps the rapid rise of this word pair can be best explained if we try to figureout its etymology, i e., the origin of the words: The little word“big” is generally arather inaccurate term Big is always relative and requires a comparison or a refer-ence to actually make sense Where there is big, there also has to be small.“Data” isjust as imprecise Those who already know a little about computer science might beable to distinguish between data, information and know-how For example: Thenumber 41 as data actually says very little However, if it is specified as an indication

of a person’s body temperature in degrees Celsius,¹ it makes more sense This mation in the right context provides the knowledge that this person may be inextreme danger (unless they’re sitting in a sauna)

infor-So the special thing about Big Data is not so much the two individual words, butactually their combination: Big Data– how big is the data that is stored about meactually? And where and how is it stored? And how is it analyzed? What benefits anddamage can I assume? How far is“big brother” from Big Data? In 2013 the scandalssurrounding the NSA (National Security Agency) and NSU (National Socialist Under-ground) caused a stir around the world– and there, too, the issue of Big Data wasinvolved: Edward Snowden courageously made public the vast amounts of data gath-ered by organizations such as the NSA and the extent to which they are stored andinterpreted The scandal surrounding the gang of murderers in the NSU showed usthat data may be at hand, but for various reasons may not be made available to theright people or appropriately correlated and connected When it comes to the media,these two scandals are certainly an important subtext of the meteoric rise of Big Data

Its fascinating inaccuracy is therefore both an advantage and a disadvantage of thisnew expression It is impossible to show the impact of Big Data² in all areas BITKOM,

1 For the sake of better readability, gender-specific differentiation will be dispensed with Gender

spe-cific terms shall apply to both genders for the sake of equality.

2 Mayer-Schönberger/Cukier: Big Data: Die Revolution, die unser Leben verändern wird 2013.

for example, calls Big Data the“raw material of the 21st century.” Again, this phor also permits both interpretations: Raw materials are the source of wealth– butraw materials have been and still are the cause of wars… Hype or hubris – both areincluded in Big Data.

meta-1.2 The First Mister Big Data?

Maybe it was just a little mishap that would go on to dramatically change the way wesee the world: An Italian scholar named Galileo Galilei had acquired a telescopeinvented by Jan Lippershey thanks to his good contacts in what is now the Nether-lands This happened about 400 years ago, in the year 1609 He looked at it and wasmarveling at the four-fold magnification of the world– when it slipped out of hishands and fell to the ground This isn’t historically documented, but it could havehappened He looked at the lenses and the construction While trying to reassemblethe telescope, he had the idea of arranging the lenses differently and perhaps also ofusing differently curved lenses After only a short time he had reconstructed a tele-scope that enabled a 33-fold magnification of the world Of course, he was also able tolook into the windows of the houses and mansions of his neighbors– but that hadlittle impact on the understanding of the world Instead, he observed the night skyover Veneto– including Jupiter with its various satellites, and of course the moon.Every night he looked at the moon and especially the shadows caused by thehills and craters on its surface He studied them with increasing excitement andbegan to sketch and chart them

When trying to explain the shadows on the moon, he began to have secondthoughts and then became certain: They could only be explained if the sun did notrevolve around the earth, but the earth revolved around the sun and the moon, inturn, around the earth

The shadows that became apparent to him not only caused him quite some nal trouble (“And yet it moves.”) – but also destroyed the Ptolemaic or geocentricview of the world and proved the heliocentric system according to Nicolaus Coperni-cus The work“de revolutionibus orbium coelestium” published in 1543 (in the year ofhis death, for safety’s sake) had highest heretical potential! And now Galileo Galileiwas actually able to prove that with some sketches of the moon?³ The millennia-oldunderstanding of the earth as the center of the world was destroyed by a few smalllenses So was Galileo thus actually the first Mr Big Data? It wasn’t until November 2,

perso-1992 that Galileo Galilei was formally rehabilitated by the Roman Catholic Church Ifand when Edward Snowden is rehabilitated remains to be seen And yet this compari-son of the two individuals may not draw a consensus everywhere

3 Recommended reading on this topic: Bertolt Brecht: Leben des Galilei: Schauspiel 1998.

1 Intro Big Data for Healthcare?  3

Source: Wikipedia

Fig 1.1: Drawing of the moon by Galileo Galilei

Even at that time a certain Venetian named Conde was supposed to have recognizedthe military potential of this invention and wanted to secure that knowledge exclu-sively for a specific purpose: He imagined that with the help of such a telescope onthe look-out of his three-master he could make a better and quicker decision as towhether attacking an approaching fleet would be worthwhile and successful– orwhether turning his ships around before they were discovered was the better move.Data analysis, knowledge and power with new technology– but the business acumen

of other Venetians at the time was greater and the profits from the sales of these

“miracle glasses” higher The telescopes therefore quickly turned out to be a real

“blockbuster,” not only for the noble houses of the world

1.3 Mr Big Data in Medicine?

If you turn a telescope around, you get a microscope So in the following years,researchers tirelessly examined all kinds of living and non-living things under

increasingly sophisticated microscopes Thus not only the world of suns, planetsand moons was unlocked, but also that of materials, cells and microbes – andmedical interest often played a dominant role.

In Berlin in 1882 Robert Koch also sketched the things he saw through his scope with increasing enthusiasm:

Fig 1.2: Drawing by Robert Koch

1 Intro Big Data for Healthcare?  5

In the cells he looked at, he saw small rods Amazingly, these were no artifactsbut actually even withstood treatment with acids The discovery of this previouslyunknown inhabitant of the earth– namely the tuberculosis pathogen – broughtRobert Koch world fame and the Nobel Prize in 1905 In order to illustrate his dis-covery to the notables of the Berlin medical society, he used something of anearly multimedia show: He built an oval ring with rails from a toy railway set on ahuge table that was especially built for that purpose in a room in the Dorotheen-strasse in Berlin (whether he borrowed the rails from his children is not known).

He then set up his microscope focused on the tuberculosis specimens on a smalltrain trailer, which he elegantly and interactively moved from one astonishedobserver to the next Let’s just hope that the mighty moustaches, the “look” of theemperor, fashionable at the time, did not get caught in the little cogs of the micro-scopes

So how does the story go? Robert Koch discovered the tuberculosis pathogen,

he invented tuberculin as an antidote and thus provided a medicine to rid humanity

of the pathogen that had probably caused the death of millions of people around theworld…

But unfortunately that story is not true Perhaps the most important and difficultpart about Big Data, then as now, is not only collecting the data, but also interpret-ing it and drawing the right conclusions from it

1.4 Mr Interpret Big Data

Thomas McKeown was a British physician and medical historian He put forward thetheory that it was not so much the medical achievements that contributed to the pro-longation of life In 2008 the highly respected journal Lancet wrote that“Tuberculo-sis lost about 75 % of its mortality from its known high before streptomycin wasavailable.” Furthermore, the tuberculin with which Robert Koch wanted to becomerich proved to be ineffective

McKeown asked the provocative question:“Does medicine matter?” Does cine actually play a role? In the self-image of most physicians it certainly does.“Doyou know the difference between a heart surgeon and God…?” The audience at a

medi-“Big Data conference” was recently asked that question “Well …,” the answer goes,

“God does not think he is a heart surgeon.”

Nobody in the 21st century would seriously try to question the many ments of medicine

achieve-However, it was many years before we were able to understand and prove thevalue of refrigerators for the health of the world’s population Neither then nor nowwould anyone seriously think of elevating Alexander Twinning, the man who com-mercially marketed refrigerators from 1834 onwards, to the Olympus of medicine

But to fully and properly collect and interpret data is something that will becomeever more important in future Maybe someday a researcher will receive the NobelPrize in Medicine for finding the right algorithm to correctly“calculate” or “antici-pate” sepsis or some other disease? Perhaps our perspective is sometimes far toonarrow– and we do not notice it? Is today’s focus on genetic data and the modifica-tions thereof (“omics”) due to the fact that it is a new field and we can calculate greatamounts of data for it– which IT companies with their great machines are of coursevery happy about and therefore often like to join forces with researchers in the questfor external funds? Or can Big Data in the field of medicine help us to better under-stand not only very specific but also more fundamental problems of health anddisease, and then to draw the appropriate conclusions?

For example, do you know what is probably the largest and most importantpublic health problem? Well, it is probably a lack of exercise: According to the

“Bewegungstudie” (survey on exercise) by TK from 2013, “Germany is sitting itselfinto sickness.” According to it, 2/3 of the population in Germany spend less thanone hour per day moving or exercising– with far-reaching consequences for theirweight, blood pressure and metabolism Will Big Data show us that the bestinvestment in health will be through cycling (individual effort) and a soundnetwork of bicycle paths (political decision)? Or do we actually already knowcertain things, but find that changing them is perceived as being a difficult task,since it not only has to include technologies but also organizational and person-nel changes?

Pretty much all publications on Big Data agree: The core competence is the correctinterpretation of data Significance, correlation and causality are not the big killjoysthey are sometimes made out to be, but actually the essential preconditions of Big

Data A historical example: Once upon a time in the Far East or the South, the ruler of

a great empire noticed that there were more deaths in a particular region than where He then had the phenomenon investigated His experts came, counted and con- cluded that there was a strikingly large number of doctors in that particular region So the ruler had all the doctors killed.

else-Admittedly, this little story is a rather drastic way of showing the difficulty ofBig Data– in medicine too: What is correlation, what is causality? Were the doctorsreally all quacks and did not know what they were doing? Or were there so many ofthem there because they were needed so much? Or did the deaths in the region havevery little to do with the medicine or the doctors?

Is it a statistical artifact? Is it often better to trust our “gut feelings” than tointerpret large amounts of data incorrectly?

1 Intro Big Data for Healthcare?  7

1.6 Mr Understand Data

Gerd Gigerenzer’s book, Gut Feelings, was the science book of the year in 2008 Are

complex and rational decisions not as good and practicable as we may think? Are nitive and analytical procedures not as successful as often claimed? The director ofthe Max Planck Institute for Educational Research emphasizes the importance ofintuition and rules of thumb– and thus of “gut feelings” – over rationality Based on

cog-a multitude of studies he wcog-as cog-able to prove thcog-at“half-knowledge” and “take thebest” are often superior to complex analytical strategies In a study he asked Germanand US-American students the question“Which city has the larger population: SanDiego or San Antonio?” The surprising result: The German students stated the rightanswer more often (San Diego) They had often never even heard of the other city–

in contrast to their American counterparts Gigerenzer assumes that somewhat (!)uninformed decisions are more likely to lead to success At the Harding Institute forRisk Literacy, he and his team explored this in the field of medicine in particular–and were able to show how low the know-how of medical professionals in this envir-onment actually is

All those who have never heard of alpha and beta errors, cannot distinguishsensibility from sensitivity, and can hardly tell the difference between absolute andrelative risk reductions will find a quiz and the“Unstatistik des Monats” (the statisti-cal absurdity of the month) on the Institute’s website as a further introduction to thetopic

1.7 Datability

The main theme of the Cebit 2014 was Big Data At the same time a further buzzwordwas created:“datability.” This is to be understood as “using large amounts of data

at high speed in a responsible and sustainable way.”

In the future, too, there will be new, previously unknown neologisms whichshould be able to address the use, uselessness, value and risk of this technology.Only the future will decide whether a comparison with radiation is appropriate

At the beginning of the 20th century people like Henri Becquerel and Marie Curiediscovered“radioactive radiation.” Since then, a broad discussion has taken place

as to how the related technology can and should be used: The use in the field ofnuclear medicine as well as radiology (CTs, MRTs) is certainly widely accepted, andthe benefits are pretty much undisputed A somewhat more disputed issue is nuclearenergy, where you will find equally vehement supporters and opponents On thiscontinuum of acceptance, the construction or use of nuclear weapons can be found

on the other side– this is where a critical distance, if not an absolute rejection, ispredominant

So with regard to today’s Big Data, are we standing at the brink of a key ortransformation technology? What is required is a more intense general and politicaldiscussion about the assessment and the use or prohibition of these technologies.Let us refer to the comparison of nuclear technology with medicine:

On the one hand, particularly in medical facilities, there are databases that

we have not yet analyzed and which are left to lie fallow, as it were Where andhow data in medicine could and should be brought together– by politics, by com-panies, and even by individuals– is what this book is all about There will be newapplications, which will surely be seen to be universally accepted Add to that themedical Big Data applications that are already happily used by some peopletoday (for example applications relating to the“quantified self”), but which arerejected by other members of the human race with the shake of the head or a fit ofrage

For this evaluation we also have to include the extreme: Today, certain software

is already subject to the same regulations as on arms exports Which basicallymeans: Applications can be used as weapons The potential for“dual use” (i e., theuse for peaceful as well as for military purposes) in medicine too, is at the core ofBig Data

1.8 Big Data in the Bundestag

The Scientific Service of the German Bundestag⁴ dealt with the topic of Big Data inNovember 2013 Several members of parliament and the public wanted to know thedefinition of the term which is currently being used so often in political debates:Big Data referred

not to an individual new technology In fact, Big Data describes a bunch of newly developed methods and technologies that enable the capture, storage and analysis of a large and arbitra- rily expandable volume of differently structured data.

For the IT industry as well as for users in business, science or public administration,Big Data has therefore become the big topic of innovation in information technology

1.9 The Three Vs of Big Data

Data is today essentially defined by three characteristics, called the“three Vs.”

1 For one, it is thevolume of data that is produced by the progressive

digitaliza-tion of virtually all aspects of modern life in unimaginably large quantities,

4 Deutscher Bundestag: Nr 37/13 (November 6, 2013), 2013.

1 Intro Big Data for Healthcare?  9

which doubles roughly every two years Thus, estimates for the year 2013 run toover 2 sextillion bytes of data being stored around the world– which, if storedand stacked on iPads would result in a 21,000 Km long wall Humanity sendsapproximately 220,000,000 e-mails around the world per minute.

2 Velocity: While in the past data was accumulated in specific intervals that

enabled you to process it little by little, today one is continuously exposed tothe flow of data due to networking and electronic communication In order to

be able to use it, the incoming information has to be taken up and analyzed atever faster rates or even in“real time.”

3 Avariety of data occurs today in particularly diverse and complexly structured

sources such as social networks, photos, videos, MP3 files, blogs, searchengines, tweets, e-mails, Internet telephony, music streaming or the sensors of

“intelligent devices.” All kinds of subjective statements in written or spokencontributions that express moods or opinions are particularly interesting forareas such as advertising, marketing or even election campaigns In order tomake the latter machine readable requires programs that can recognize judg-mental statements or even emotions about products, brands, etc., which is tech-nically very challenging

According to estimations by McKinsey Global Institute, the use of Big Data couldcreate annual efficiency and quality increases to the value of around EUR 222 billionfor the US-American health service alone, and approx EUR 250 billion for the entirepublic sector in Europe

The exceptional thing about Big Data analyses is particularly apparent in thenew quality of results from the combination of data that has so far not been related

to any other data

1.10 Can Medicine Learn from Other Industries?

The question is whether anything can be learned from other industries, and if so,what: According to Mayer-Schönberger, as many as two-thirds of the shares on USmarkets are today traded through computers and their algorithms– in milliseconds!

So can we expect future scenarios in which we would do better to ask Dr Computerthan our general practitioner?

To a certain extent, global digitization is far ahead of medicine: In the year

2000, only 25 % of the globally available information was said to be digital In theyear 2013,⁵ however, only 2 % was said to be not digital “If this information werestored in books, it would cover the entire area of the United States with a stack that

5 Mayer-Schönberger/Cukier: Big Data: Die Revolution, die unser Leben verändern wird 2013.

was 52 books high”⁶ – striking, often unquestioned or unquestionable comparisonsare also a part of Big Data More of it?

The great library of Alexandria was built in Egypt during the reign of Ptolemy II,and was supposed to represent the entire world’s knowledge Today we live in aworld in which 320 times the amount of data contained in the ancient library is allo-cated to each and every human being…

Eight million books are said to have been printed between the years 1453 and

1503 Therefore more data was produced in those 50 years than in the previous 1200years, from the founding of Constantinople Today we produce that amount of data

in three days alone…

The fact that quantity can also have a new form of quality is described by Schönberger using the example of a horse: An oil painting, a photo of a horse and therapid scrolling of photos with more than 24 frames per second– this changes not onlythe mere number of images, but the essence itself Today this example can be easilytransferred to medicine: From anatomical drawings, to ultrasound images, to ultra-sound films– and to 3D ultrasound animation…

Mayer-Amazon’s worldwide success is also due to the way that people who havepurchased a certain product will automatically be shown other products thatwere bought by people who purchased the same product In the early years of thecorporation, an internal study found that this automated assignment of recom-mendations is superior even to an editorial team of booksellers who make theirown recommendations based on buyer profiles– and, of course, it is also faster,cheaper and more scalable Amazon may not actually know why someone whoreads Goethe will also listen to Beethoven, but more than a third of Amazon’sturnover is today said to be generated through recommendations based on algo-rithms

So do we have to analyze and interpret data differently in medicine? Is it vable that during the course of a disease we do not actually understand why some-thing happens, but that it is likely to happen and that we can therefore respond to itbetter?

concei-Should we, as proposed by the author of Big Data,“possibly give up some racy in order to detect a general trend”?

accu-2 Big Data Framework in Medicine

The Big Data Framework summarizes the key elements for applications in medicine.The framework consists of technical and non-industry-related elements, amedical, a legal and an organizational dimension

6 Mayer-Schönberger/Cukier: Big Data: Die Revolution, die unser Leben verändern wird 2013.

1 Intro Big Data for Healthcare?  11

Analysis, Visualization, Decision Support

Data Management (Integration and Sharing)

municipalities,

WHO…)

Insurance Data

(health insurance companies, insurances, public institutions )

Research Data

(bio banks, clinical trials, PubMed…)

Individual Data Generated by Patients

(wellness, nutrition )

Non Classical Healthcare Players

(social networks, telco, retail…)

Governance (data protection, IT security, informational self-determination)

Source: Author's illustration.

Fig 1.3: Big Data Framework in Medicine

The following provides a brief bottom-to-top explanation of the areas and illustratesthem by examples:

– The area of governance in particular refers to basic organizational and legalconditions– these may relate to the individual company or institution, but also

to basic national or international frameworks

– Data sources in the medical context are manifold: For one thing, they constitutethe medical data that is primarily found in hospitals In addition, these can becomplemented by or correlated with data from public institutions– including, forexample, municipalities, public health departments, ministries or internationalorganizations such as the WHO (World Health Organization) Depending on thehealth system, data about the insured persons (for example when claiming bene-fits) can be integrated with the health insurance companies, other insurances

or even private entities – such as HMOs (Health Maintenance Organizations).Research data– with its often very specific characteristics and collected underspecial ethical and data protection requirements– is another category Individualdata generated through patients (pain journal, data entered through sensors…) isgenerally still not a part of the health system today, but will significantly increase

in importance in the near future Still practically not involved today are the called “non-traditional players.” These include data from social networks(e.g., Facebook), data from mobile operators or even retailers (“consultation willtake place in the supermarket”)⁷ – and even the German postal service is said to

so-be thinking about using mail carriers for the home delivery of medication…

7 Online: https://www.trendmonitor.biz, [accessed on: July 30, 2014].

– Infrastructure, which primarily describes the physical conditions for the port, connection and storage of data.

trans-– Data collection and archiving is particularly significant in medicine– first in terms

of data security (as recommended by the Federal Office for Information Security–BSI) but also in terms of the legal requirements for long-term archiving.⁸

– The term “data management” includes not only the integration of data (inaccordance with existing standards such as DICOM IHE or HL7), but also theunderstanding of role assignment and the control of data access

– The actual merging of data in order to present it to various user groups and toanalyze it is therefore only the“surface.” Data can “merely” be shown or if not,then specific constellations of data can cause direct consequences

This framework is not intended to replace existing models such as the OSI model.OSI (Open Systems Interconnection Model) is a reference model for network proto-cols as a layered architecture with seven so-called layers The“Big Data Framework

in Medicine” is a framework that identifies the key dimensions of Big Data in cine that go far beyond a purely technical point of view

medi-2.1 Fields of Action in Medicine

The fields of action for Big Data applications are already multifaceted today– andfuture scenarios will also be described in this book that were almost unimaginablenot long ago

Decision Support Support)

Disease Management

Public Health Monitoring

Medical Performance Optimization

Consumer Behavior

Information Exchange (Patient to Patient p2p)

Information Exchange (Patient to Patient p2p) Prevention of

Misuse

Optimized Clinical Studies Product

Quicker Implementation

of Evidence

Control of Regulations

Prognosis of Dieseases Governance (data protection, IT security, informational self-determination)

Source: Author's illustration.

Fig 1.4: Fields of Action of Big Data in Medicine

8 GMDS: Dokumentenmanagement, digitale Archivierung und elektronische Signaturen im

Gesund-heitswesen 2012.

The main scenarios are described here and outlined using examples– of courseother overlaps of dimensions are also possible.

The different participants are shown on the left side of the diagram The boxesand scenarios indicate the focus of areas– without excluding other cross-references.– Health Education and Information refers to all areas that compile or upgrade

data for different participants, sometimes for the first time (for example the design of disease management programs as well as quantified self approaches).– The analysis of the inputs (resources) enables new forms of resource control

re-and thus output (medical quality, effectiveness) This enables organizational proaches, for example (pay per performance) or the design of new forms of care.– An example ofPublic Health Monitoring are the faster and more targeted health

ap-measures based on health trends as well as individual health-promoting activities.– The rapid feedback of subjective and objective data can also contribute toward

product development – such as new services in the field of disease care as

well as medical devices in the AAL (ambient assisted health) environment.– Theprognosis of diseases with an “individual health analytics” approach is an

example of how a better understanding of medical conditions can be achieved

by the integration of data from the living environment (empowerment)

– Data integration and analysis enable faster and more targeted forms of tion as well as an adaptation of campaigns to current challenges or a response

preven-to acute threats, such as epidemics

– For example,disease management programs can be offered in a more

indivi-dualized way in future

– Recruitment forclinical trials as well as data simulation can enable new forms of

study faster, and connect bed and bench (clinic and research laboratory) better.– New services and products can be designed for the extended healthcare

market (nutrition, fitness, wellness…)

– Medical performance optimization can enable better implementation of

current guidelines (“coach the doctor”) or transparent outcome measurement.– Exchange of information from patient to patient: The first and most famous

example of this is http://www.patientslikeme.com/– an internet platform onwhich patients can find other patients with similar symptoms and experiences.⁹– Communication describes all the processes and scenarios that help to over-

come the silos of the institutions, and enable new forms of planning and mentation of health services

imple-– Adherence means scenarios in which a person’s behavior, i e., the intake of

medication, a diet regime and/or a change in lifestyle, corresponds better to therecommendations agreed to with the therapist

9 The founder of Patientslikeme recently made a plea to all individuals to donate their personal

health data The partners of the US company include academic institutions as well as pharmaceutical companies – that would certainly also be interested in the patient data.

– New fields of business will arise for providers who so far do not belong to the

group of traditional players– e.g., a postal service that takes on care services,supermarkets that are part of a diet program, or telecommunication companiesthat provide telemedicine services

– New forms of digital decision support include, for example, not only new

kinds of visualizations, but also the improved integration of genetic, clinicaland research data and the provision of activity algorithms

– Big Data scenarios also enable better protection againstmedical malpractice.

– Control scenarios will be able to incorporate new indicators that address

indi-vidual or geographical components more effectively

– Faster implementation of evidence can be an important focus of Big Data

applications

– Exchange of information between doctor and patient can include

telemedi-cal services, or real-time data analysis, even via sensors Here there is morethan one possibility: scenarios that have a stronger procedural focus (e.g., dia-betes monitoring) or applications that accumulate and analyze existing data

The following diagram shows the potential and the scenarios of Big Data arisingfrom a time perspective: New models can be developed on three coordinates:

Temporal: yesterday, today, tomorrow…

Vertical: Clinic – Research

Horizontal:

Outpatient – Inpatient

Source: Author's illustration.

Fig 1.5: Dimensions of Big Data

1 Horizontal: This could include the better integration of data along the

treat-ment chain This means not only the combination of outpatient, inpatient and

rehabilitative treatment, but perhaps also information from everyday life thatcould be included in a sensible way.

2 Vertical: This entails the deeper integration of databases, such as the better

connection of administrative, clinical and research-related data Anotherexample is the “open data initiative” of the British Medical Journal The BMJwill in future only publish articles that have been registered in advance, handlethe results/data in a transparent way and disclose them accordingly And weare not talking about“peanuts”: A large part of clinical studies have never beenregistered or published – according to the BMJ approximately 50 % of allstudies, with a significant number of unknown cases More than 99 % of thedata collected for research purposes (with considerable effort and commitmentand possibly risks to patients) is no longer available after its publication If wehad access to all that data, how much more would we know?

3 Temporary: Big Data analyses not only encompass a backward glance, but also

enable the real-time monitoring of business processes The next step is from“whathappens” to “why something happens.” This is where further analyses may providenew insights A particularly exciting area is where future scenarios can be simulated

or maybe even“predicted”: To monitor a (potentially) complicated course of therapyearlier on– or to even take appropriate measures – might become a high point ofBig Data A look into the future is particularly fascinating, and it is where longings,

visions and desires come together: What will happen, what risk profiles are there?

And how does this permit earlier preventive intervention, a change/expansion inclinical diagnostics and the adaptation of therapies? After all, health does not takeplace in hospitals, but in real life So how can we integrate that area better?

1 Machine based

2 Hypothesis based

Examples

5 Evaluation of clinical pathways

8 Identification of inappropriate medication

6 Evaluation of drug efficacy based on real world data

Risk stratification/patient identification for integrated care programs

3 4 6 2 7

High

High Low

Systematic identification of obsolete drugs usage 10

8

10

Personal health records 11

Source: McKinsey & Company: McKinsey Big Data Value Demonstration Team.

Fig 1.6: The aspiration: Better understanding of what happened, what will happen

In a whitepaper called “The Big Data Revolution in Healthcare” (2013), McKinseyand Company summarized the scenarios that will be applicable in medicine.

So where are we today? The reporting in hospitals today is mainly looking: What happened in the past? In the best case it might be: What is happeningright now? In the area of data mining and evaluation in terms of a real-time perfor-mance evaluation, initial projects already exist, and HMOs are working on detailedrisk stratifications of patient populations

backward-2.3 The Example of Google Flu Trends

A particularly prominent example is the discussion about“Google Flu Trends” – aweb service operated by Google, which aims to correlate the entry of certain searchterms with the regional occurrence of diseases– for example the flu Without addi-tional instruments, the use of the search engine is meant to recognize diseasetrends

Google employees themselves published an article titled “Detecting influenzaepidemics using search engine query data” in the scientific journal Nature: 50million search terms were analyzed per week between 2003 and 2008.¹⁰ This wasbased on 42 prediction parameters that were identified from the data The authorswere pretty euphoric and stated that they can predict influenza-like symptoms accu-rately within a day

Because the relative frequency of certain queries is highly correlated with the percentage of physician visits in which a patient presents with influenza-like symptoms, we can accurately estimate the current level of weekly influenza activity in each region of the United States, with

a reporting lag of about one day.

Google was celebrated: At last important evidence had been found that there was abenefit to be had from Big Data in medicine

But little by little doubt and skepticism crept in, which resulted in a detailed(“final final”) summary of the discussion in the academic journal Science.¹¹ Under the

title“Big Data and The Parable of Google Flu: Traps in Big Data Analysis,” the casewas re-analyzed and summarized: Google had predicted more than twice as manycases of infection than were actually later confirmed by the CDC (Centers for DiseaseControl and Prevention), the national health authority of the United States! The authors

in Science demonstrated that there may be many examples that show how the analysis

10 Brammer/Brilliant/Ginsberg et al.: Detecting influenza epidemics using search engine query data.

In: Nature 457, February 19, 2009 Online: http://www.nature.com/nature/journal/v457/n7232/full/ nature07634.html, [accessed on: August 6, 2014].

11 Lazer/Kennedy/King/Vespignani: The Parable of Google Flu: Traps in Big Data Analysis In:

Science 343/2014, pp 1203–1205.

of search engines and social networks could predict certain things.¹² However, in trast to such propositions, we are in fact far from being able to replace establishedmethods The main points of criticism are:

con-– The quantity of data cannot replace the quality of data

– Search engines or social networks are not designed for medical data collection.People“googling” a particular disease, will not necessarily have that diseasethemselves (but may be looking out of pure interest or for someone else).– The construction mechanisms of studies, whether epidemiological or clinical,such as for measuring validity or reliability, cannot be replaced by pure volume.– The search algorithms were not constant and were adapted by Google itself(“blue team issues”)

– So-called“red team attacks” may be unlikely, but cannot be ruled out in such aconstruction: This refers to “data attacks” attempting to manipulate publicopinion by entering certain search terms artificially and with manipulatoryintent (whether by human hand or computer)

– Lack of transparency and the inability of independent repetition (replication).– The combination of large amounts of data with small search terms inevitablyleads to a problematic marriage of the two, with statistical consequences

Nevertheless, the authors claimed that the data collected by Google was definitely ofvalue, though they admitted that it must be combined with other data sources, such

as those made available by health authorities

2.4 Big Data Adoption Model in Medicine

Different approaches exist on how to measure the maturity of an organization when

it comes to Big Data From these one can then develop something of a“roadmap”for future planning

The key parameters are:

– Data collection: How is data currently collected, and to what extent has the

level of digitization progressed?

– Data sharing: Is there a possibility of data exchange within or outside of an

organization? Are standards available and maintained for that purpose?

– Data analytics: At what stage of development is the current data analysis

(“data warehouse”) and what percentage of employees can use specific data

12 “The problems we identify are not limited to GFT Research on whether search or social media can

predict x has become commonplace and is often put in sharp contrast with traditional methods and hypotheses Although these studies have shown the value of these data, we are far from a place where they can supplant more traditional methods or theories ”

visualizations? What specific algorithms and modeling of data are there that gobeyond traditional “reporting”? Are there any IT-based tools to develop fore-casts and models?

A concrete example is the Big Data maturity level guide,¹³ which includes a faceted plurality of dimensions, and analyzes the maturity of the organizationaccording to the stage of development

multi-In 2013 Verhej¹⁴ examined the process of Big Data adoption on the basis of cific case examples He differentiates five core areas: strategy, knowledge develop-ment, piloting, implementation and fine tuning In particular, he emphasizes the

spe-“business case” – because not everything that is possible is sensible or providesadded value In the health care industry the“roadmap” issue is currently becomingparticularly important, since only with an overview can the incredible heterogeneity

of systems be overcome and at the same time the benefits be made transparent

Old Wine in New Bottles?

In a hospital context– this area exemplifies the entire health care industry – data istoday already generated, collected and evaluated So what is new and explains thesuccess of Big Data?

This is where three trends come together and converge in Big Data:

1 Technologically there are new ways of processing huge amounts of data in real

time, for example through in-memory database technology (processing largeamounts of data in the main memory, SAP HANA) and NoSQL databases (“Notonly SQL” databases), which are optimized for dealing with unstructured dataand provide query interfaces via SQL– for example the prominent open sourcesolution Hadoop, a highly scalable, highly parallel data storage and data pro-cessing technology That was simply unimaginable a few years ago

2 In addition, every hospital has something you could call “silo experience.”

Data is distributed, buried and even hidden in so many different, separatesystems, meaning that there is a need and even a deep desire to merge every-thing One could also call it“a treasure that lies hidden in fragments.” So it isalso about bringing together existing data in“real time” and possibly develop-ing new business models from that What would Amazon or Facebook do withthe data generated in a university hospital? This idea may be fascinating forsome– but for others it’s a nightmare…

13 Halper/Krishnan: TDWI Big Data Maturity Model Guide 2014.

14 Verhej: The Process of Big Data Solution Adoption 2013.

3 These new technologies lead tonew applications An example of this might be

knowledge-based systems, which can really “coach” a treatment process andmerge large amounts of information for that purpose Until now, IT in hospitals,particularly in the clinical field, has been primarily perceived as a data servantand not as a coach, who can provide support whenever necessary The word BigData brings these streams together“crisply.”

3.1 The Difference to Classic Clinical Analysis (Old Wine in

New Bottles?)

Big Data is centrally connected with the integration of different views: Medical ling also has to be seen from a quality perspective– not only a monetary one On theother hand, quality assurance and evidence-based medicine also have to be consid-

control-ered in economic terms The traditional separation between management (finance), medical (hospital) and potentially also research has to be abolished The call to do so

may be“old wine.” Today we have no more new bottles, but we do have a complete,ultra-modern filling station And what’s more, it also enables us to provide the correctbrew for a variety of different tastes So– not just old wine for everyone, but evencocktails for those who want them…

3.2 Is Big Data in the Hands of Medical Controlling Degenerating

to a DRG Optimization Technology?

If software is used in a first phase to optimize revenue and improve the management

of a hospital– that is certainly a good start But it is only a first step Modern tals, such as the Charité, understand that they need a strategic platform for inte-grated data analysis In this case medical controlling is surely“low hanging fruit”with which to devise a process-cost-based hospital in the future If we think of thevision of“outcome based payment,” it will only be possible if data at the micro level(the patient), meso level (the department) and macro level (the hospital) are merged

hospi-in order to provide any khospi-ind of control at all…

But the real problem in today’s hospitals is much more banal: Even in sity hospitals the majority of documentation still takes place in part on paper– andwithout digitization that cannot be computed, even with Big Data DRG optimiza-tion has not only an inward but also an outward dimension: Being able to simulateprice models based on new assumptions but with “old” cases when negotiatingbudgets with health insurance companies can quickly yield millions of Euros forthe hospitals…

univer-3.3 Are Hospitals Prepared for Big Data?

Generally: Yes and no! IT is often not given the central strategic importance that itshould have This is something medical information scientists have been lamenting foryears, but have not managed to do much about, as they often have difficulty explainingthe strategic added value in the words of the executive management How many hospi-tals have a Chief Information Officer (CIO) who is a member of the executive board?Very few How many Chief Medical Information Officers (CMIOs )– a person who strate-gically and explicitly“takes care” of medical IT needs – are there in German-speakinghospitals? Very few… Big Data is not about which server to buy – but about how I canhandle my central nervous system in the medical“knowledge industry,” i e., how Ican measure and act This message and this understanding have not yet arrived inmost hospitals CXXOs– this is not a title on a business card, but a top managementunderstanding that I should not leave this decision to my IT department alone Ne-vertheless, many think Big Data is a central issue for the future Some hospitals arealready beginning to develop strategies in this area, but we are just getting started.Furthermore, experts around the world agree: The problem is not the technol-ogy, but that the people/staff possess enough statistical know-how to deal with it.Whoever cannot distinguish absolute risk from relative risk and thinks that correla-tion equates to causation should stay away from Big Data

3.4 Are Internal Hospital Big Data Analyses Data Protection Cliffs, and if so, Where Exactly?

As with all processes– whether with or without IT – privacy and personal rights must

be protected However, if privacy (or sometimes the pretext of privacy) preventsimportant insights with which to improve patient treatment, then it has to be scruti-nized– because otherwise privacy protection will mutate to perpetrator protectionand become a common excuse to prevent much needed transparency On the otherhand, Big Data sounds similar to“big brother” – which is a particularly sensitiveissue, given the discussion about the NSA and misuse of data by intelligence services

We are only at the beginning of a political (note well!), not of a technical cussion about Big Data– and thus one of the most important raw materials in theglobalized economy Again, this invokes comparison with the discussion about thebenefits of nuclear energy in the 1940 s and 50s: Today nuclear weapons are scorned,nuclear power plants are a controversial issue, and nuclear medicine departments arewelcome– but we are generally talking about the same technology

dis-Where, when and to whom can I give my data and thus“my innermost self,”and which organizations and companies may use what? This is where we need clearrules and laws and no unresolved big gray areas Today in Germany, we have notonly a Federal Data Protection Act, but also local forms of federal data protection

Regulation of data protection on a European level, such as the European ment’s draft for data protection regulation, EU-DSGVO, will be necessary Thisshould make a contribution toward unifying data protection legislation in Europe.¹⁵The protection of privacy and the personal data of citizens is a general socialconcern in Europe This applies to all areas of life– whereby, in this context, thesubject of Big Data is particularly sensitive The fundamental right to the guaran-tee of the confidentiality and integrity of information technology systems (collo-quially referred to as basic IT law, fundamental computer rights or fundamentalrights to digital privacy) is a fundamental right applicable in the Federal Republic

Parlia-of Germany, which primarily serves the protection Parlia-of personal data stored and cessed in information systems This right is not specifically mentioned in the basiclaw, but was thus specified as a special form of the general right of personality in

pro-2008 by the Federal Constitutional Court or derived from existing fundamentallegal provisions respectively.¹⁶

Similarly, a discussion regarding informational self-determination and the

“right not to know.” This is the point at which we reach a complex ethical sion, which we have to face

discus-4 Personalized Medicine and Big Data

To provide each patient with an individual therapy that acts painlessly, quickly andeasily– for some that is a vision, for many others it might sound delusional Oftenpersonalized medicine, particularly in the context of Big Data, is understood to meanthe connection of genetic code and possible healing options Hardly any other area ofmedicine has seen such an increase (“explosion”) of data in the last five years, withthe corresponding faster and cheaper analysis of the data Therefore, the hype aboutBig Data and genetics can also be explained through these new developments.The German Cancer Research Center announced in May 2014:

Extensive genetic analyses of cancer cells have shown: Every tumor is different and every cancer patient must be treated individually The National Center for Tumor Diseases (NCT) Hei- delberg wishes to meet these demands: From 2015 onwards, patients at the NCT will be offered

a genetic analysis of their cancer cells according to which an individual therapy can be mended.¹ ⁷

recom-15 Press release and statement by TMF, July 2014: “Die Daten der Bürger schützen, biomedizinische

Forschung ermöglichen: Medizinische Forscher kommentieren den Entwurf einer europäischen Datenschutz-Grundverordnung ”

16 Wikipedia: Grundrecht auf Gewährleistung der Vertraulichkeit und Integrität

informationstech-nischer Systeme Online: http://de.wikipedia.org/wiki/IT-Grundrecht, [accessed on: August 1, 2014].

17 Press release by DKFZ:

http://www.dkfz.de/de/presse/pressemitteilungen/2014/dkfz-pm-14-24-Individualisierte-Krebsmedizin-fuer-jeden-Patienten.php

Source: Author's illustration.

Fig 1.7: The“Data explosion”

So are we at the brink of a break-through of“personalized medicine” in all areas?

4.1 The (Genetic) Human in Numbers

What does the human (or preferably the parameterization in biology) mean innumbers?

The human genome (genomic DNA) has a length of 2 x 3.2 Mb with mately 19,000 to 26,000 protein-coding genes, 1500 genes for micro-RNAs, some

approxi-8900 pseudogenes and approximately 290 disease genes for monogenic (Mendelian)genetic diseases

The numbers for a transcriptome of a human (messenger RNA):

– Number of translated proteins in a cell: approx 5000 to 10,000

– Number of mRNA molecules per cell: approx 15,000

– Average length of an mRNA: 1000‒1500 nucleotides

– Average half-life of an mRNA: 10 hours

– Length of a processed microRNA: 18‒24 nucleotides

The human proteome (proteins and peptides):

– Number of protein copies of an mRNA molecule: approx 300‒600

– Average length of a protein: 375 amino acids

– Longest protein (titin): 33,423 amino acids

– Average half-life of protein: 6.9 hours

– Number of copies of a protein per cell: 18,000 (median)

Number of human tissues (cell types):

– Different types of tissues in the body: approx 230

– Organ systems in humans: 22 (categories of Human Phenotype Ontology HPO)– Number of cells in a human body 3.7 ± 0.8 x 1013

Human diseases:

– Monogenic (Mendelian) genetic diseases: approx 9000

– Monogenic (Mendelian) genetic diseases with known disease genes: approx

3800 caused by mutations in one of the 2850 disease genes

Bear in mind, this is merely the primary genetic view of a human Furthermore, the

International Classification of Diseases (ICD) as well as the Diagnostic and StatisticalManual of Mental Disorders (DSM) also include tens of thousands of groups andsub-groups of diseases

In order to understand the complexity of genetics, a visit to the Encode Projectsite¹⁸ is strongly recommended to get an impression of the diversity in this area.ENCODE (which stands for ENCyclopedia Of DNA Elements) is a research projectthat was initiated in September 2003 by the US National Human Genome ResearchInstitute (NHGRI) The project aims to identify and characterize all functional elements

of the human genome as well as the transcriptome

They recommend that visitors play with the Human Genome Browser, whichenables you to zoom into different gene sequences online.¹⁹

Some people may quickly shy away from this topic, since the complexity is fusing and unnerving

con-At this point I would like to include a personal note: When I look at these pages, I detect a sense of humility in the face of creation I get the impression that we can probably see only the tip of the iceberg in 2014… Although there are scientists who claim that a genetic terra incognita no longer

exists But is that so? If I shred the score of a symphony by Beethoven and analyze the number of notes, clefs and trebles, can I then understand the essence of the music? The causality, from genome to peptide, to active ingredient, to action and activity of a human being is not linear and not unidirectional – ”Hope is the dawn on a stormy night.”²⁰

In terms of“Big Data and personalized medicine,” there are many experts today –and probably even more, who think they are experts

18 http://genome.ucsc.edu/ENCODE/

19 The following link provides a depiction of a gene sequence: http://genome.ucsc.edu/cgi-bin/

hgTracks?db=hg19&position=chr21 zE66eZH16J4V, [accessed on: August 5, 2014].

%3A33031589-33041578&hgsid=383855821_U1XsDAYLsaHrj1AO-20 Goethe: Triumph der Empfindsamkeit, IV.

Therefore, I would like to let a“real” expert have his say An expert who hasworked in the Encode Project for years Ewan Birney is a computational biologistand Deputy Director of the European Bioinformatics Institute in Hinxton, UK, which

is part of the European Molecular Biology Laboratory (EMBL) Birney coordinatedthe data analyses of the Encode project

In a ZEIT interview by the title of“The Genome is a Jungle Full of Strange tures”²¹ in September 2012, he provided some deep insights into his work

Crea-DIE ZEIT: The Human Genome Project deciphered our genome and explored the human genes ten years ago Now you and 440 colleagues have worked on

it for another five years in the Encode Project Why, exactly?

Ewan Birney: It’s actually a part of our job to bring the search for the genes to anend, even if many people think that that already happened long ago But the genesaccount for only a tiny part of the genetic information Encode’s major goal was tofind out what all the rest of a genome is actually good for– all the non-coding DNA,which derogatorily was called junk

ZEIT: You seem to have discovered quite a bit …

Birney: We will post at least 40 publications in three professional journals online inone fell swoop Thirty of these are linked by a matrix so that readers can track andcross-reference every aspect Nothing like that has ever been done before

ZEIT: And what is it you found in the genome?

Birney: It is full of surprises There is far more going on there than we ever expected.The genetic material is full of activity

ZEIT: So we have to bury the idea that our genome consists mostly of garbage?

Birney: That’s correct Junk DNA was never a particularly appropriate metaphor, ifyou ask me I find“the dark matter of the genome” much better

ZEIT: And how much dark matter will still remain in the genetic makeup after Encode?

Birney: It is difficult to say: We understand some parts, but don’t understand others.There are too many functional levels in a genome But let me put it this way: Wehave now assigned 80 per cent of all hereditary factors to a certain biological activ-ity Of these 1.2 per cent code for all the proteins in a body, but another 20 per centare for controlling these genes

ZEIT: Alright, so do we now understand how the genetic makeup functions?

Birney: Unfortunately not I wish we did

ZEIT: But we’re looking at a kind of circuit diagram for our genetic makeup?

Birney: That is a very good analogy Only one cannot really say that our genomelooks clean and tidy The unexplored wilderness we encountered– that was a realsurprise for me The genetic material is a jungle full of strange creatures It’s hard to

21 Online: http://www.zeit.de/2012/37/Encode-Projekt-Birney, [accessed on: 5.8.2014].

believe how closely packed it is with information! We are now in the situation of

an electrician who has to check the electrical system in an old house and notices: allthe walls, ceilings and floors are covered with light switches We have to find outhow all of these switches are connected with the light, heating and equipment inthe rooms

ZEIT: And what do these genetic switches do in our body?

Birney: Take the cells of hair roots, for example They activate genes that are sible for the color pigments in the hair The liver cells, on the other hand, formalcohol dehydrogenase, the enzyme that breaks down alcohol…

ZEIT: Conclusion: Up to now, we were aware of more than 20,000 genes for proteins Now you have added millions of switches that control a cell’s biology

in a variety of combinations And all of that can also be different from person

to person Is there any hope left that we will actually understand it all one day?

Birney: We will probably need the entire 21st century for it But the good news is:You do not have to understand everything in order to make progress in medicine.For example, many of the genetic switches are located in areas of the genome thatare associated with the most widespread diseases Many of these diseases– dia-betes, bowel inflammation and the like– will be caused by errors in the switches

We can presume such switch effects for 400 diseases

4.2 Personalized Medicine?

Personalized medicine, also called individualized medicine, aims to treat each andevery patient while taking into account individual conditions that go beyond functionaldisease diagnosis The term“personalized medicine” as outlined is often used particu-larly in connection with pharmacotherapy/biomarkers and/or“genetic therapy.”However, this use of the term “personalized/individualized medicine” in areduced, biological interpretation is somewhat controversial The Federal Center forHealth Education (BZgA) emphasizes in its main definitions that the term“personalizedmedicine” is misleading in its contextual meaning in so far as “the personal side of aperson, i e., his capability of reflection and self-determination is primarily not meant atall, but that fundamental biological structures and processes are emphasized.” Thechairman of the central ethics commission of the German Medical Association, UrbanWiesing, was critical:“Personal characteristics manifest themselves not on a molecular,

but on a personal level.” Heiner Raspe from the Center for Population Medicine andHealth Care Research (ZBV) at the University of Luebeck argues that the term“persona-lized medicine” is too one-sided in the sense of pharmacogenomically based therapy;

he specifies that in addition to biomarkers there are also“psycho markers” and “sociomarkers,” which also deserve to be considered in medical therapy, as selected exam-ples have shown The Bundestag committee on education, research and technologyimpact assessment proposes the term“stratified medicine,” which is increasingly used

in international literature.²²

However, the terms“personalized” and “individualized” suggest a therapy that

is precisely tailored to the individual being treated But this is not normally the case

in modern day medicine and also not meant Basically, today all diagnostic andtherapeutic measures should orient themselves not only on the best available evi-dence, but also on the individual characteristics of a patient, and, as long as appro-priate, the patient’s wishes, in the spirit of patient autonomy

Techniker Krankenkasse²³ (technicians’ health insurance company) suggests:

In general, stratification is carried out with the help of biomarkers That is why the expression

“biomarker-based medicine” would be possible as a synonym for stratified medicine This is what is usually meant in today ’s medicine in the context of “personalized medicine.” If it is to

be clarified that the stratifying medicine is oriented on inherited genetic characteristics, the term “genome-based medicine” can also be used.

Conventional medicine according

to the one drug fits all principle

• Autologous cell therapies

• Active personalized immune therapy (tumor vaccines)

• Individual medical products (rapid prototyping)

22 Quoted according to: Wikipedia: Personalisierte Medizin Online: http://de.wikipedia.org/wiki/

Personalisierte_Medizin, [accessed on: August 6, 2014].

23 TK: Innovationsreport 2014 Online: http://www.tk.de/centaurus/servlet/contentblob/641170/

Datei/121104/Innovationsreport_2014_Langfassung.pdf, [accessed on: August 6, 2014].

according to the rapid prototyping principle In addition, APVACs (Active lized Vaccines) must also be mentioned in the context of individualized medicine,even if they have not yet arrived in everyday clinical practice This refers to a vacci-nation against cancer With APVACs the composition of a vaccine (especially tumorvaccines) is determined by previous biomarker tests The peptides necessary for theformation of the vaccine are then derived from a peptide warehouse, for example.However, such active personalized medicine is still in the development stage and isnot currently used in general clinical practice Moreover, one can argue aboutwhether the terms“personalized” and “individualized” go too far, because it is notthe person or the personality or individuality of a person that is addressed In fact,only the biomarkers of the respective people are meant.

Persona-Biomarkers are parameters with which a property can be objectively measured inorder to use it as an indicator for normal biological or pathological processes or phar-macological responses to a therapeutic intervention Biomarkers can be differentiatedinto prognostic, predictive and pharmacokinetic/pharmacodynamic biomarkers.According to TK, there are currently 27 substances available on the Germanmarket, for which a biomarker test is to be carried out according to the respectivescientific information before using the drug In most cases, the use of the drug isindicated only if the diagnostic test showed a positive result before treatment andthe biomarker, i e., a certain molecular change in the tumor cells, is present

4.3 Statement by the German Ethics Council

In April 2013, the German Ethics Council released a statement titled“The Future ofGenetic Diagnostics – From Research to Clinical Application.” For this purpose,information was provided on the latest technical methods of genetic diagnosis andits use in medical practice during a public hearing

“In recent years, the methods of genetic analysis have been developed rapidly.The new applications are designed to improve the explanation of the causes ofdisease as well as risk forecasts, and to contribute to new therapeutic approaches.However, it remains to be seen how quickly and to what extent they will find theirway into clinical practice.” The Ethics Council stressed that “the question of thequality of life should not be reduced to medical or genetic findings.”…

“With multifactorial diseases, it is not appropriate to constantly address geneticvariances as (potentially harmful) mutations Often these are actually polymorph-isms that are widespread in the population Their possible influence on disease andhealth is only achieved in the complex context of other genetic, epigenetic andenvironmental factors.”

After a period of high-flying expectations, the assessment of a direct clinicalapplication of the results of genome-wide association studies has, for now, givenway to disillusionment The irreversible conceptual difficulty is that for mulitfacto-

rially influenced characteristics, a large number of gene loci and an even greaternumber of interactions between these gene loci come into question This inevitablyleads to a tendency to statistical overfitting of the correlations, where random corre-lations between DNA sequence and phenotype are interpreted to be the putativecause Likewise, underfitting can also occur, where actually relevant genes or inter-actions between multiple genes are not or not properly recorded and thereforeescape identification.

4.4 23 and me?

“23 and me,” a Big Data start-up from the United States, certainly provides the bestknown genetic test kit As a direct-to-consumer test (DTC test), the company offered

to“analyze the individual genome” via the Internet with the help of a few milliliters

of saliva In addition to being tested for the probabilities for a variety of diseases,you can have your genetic closeness to certain animals (such as cow or mouse) ana-lyzed for USD 99.00 The company claims to have analyzed the genes of 450,000customers to this day The saliva samples were said to have been tested for about

200 genetic diseases and 99 other predispositions

In Big Data circles of the time, it was“hip” to ask “have you already or haveyou not yet…?”

The marketing machine promised much in 2009, including the“genetic tain of youth.”²⁴

foun-In 2013, the United States Food and Drug Association (FDA) banned the companyfrom selling the test, although the FDA stressed that there had been a dialogue formany years and it had tried to support the provision of legally compliant offers

However, even after these many interactions with 23andMe, we still do not have any assurance that the firm has analytically or clinically validated the PGS (Personal Genome Service) for its intended uses, which have expanded from the uses that the firm identified in its submis- sions.² ⁵

24 The following website provides further information as well as depictions: ScienceBlogs: 23andMe

offers free genome scans to 4,500 senior athletes, seeking genetic fountain of youth Online: http:// scienceblogs.com/geneticfuture/2009/08/12/23andme-doing-free-genetic-tes/, [accessed on: August

6, 2014] http://www.google.de/imgres?imgurl=http%3A%2F%2Fscienceblogs.com%2Fgeneticfuture

%2Fwp-content%2Fblogs.dir%2F274 %2Ffiles%2F2012 %2F04 %2Fi-ce6030d80cf23822cb0cea49d86a 454c-palo-alto-online_23andme-senior-athletes-ad.jpg&imgrefurl=http%3A%2F%2Fscienceblogs.com

%2Fgeneticfuture%2F2009 %2F08 %2F12 nid=Ga9SEySaOG5WSM%3A&zoom=1&docid=Z0kQx3OevDC5IM&ei=3t3ZU7S7D8an4gTa3oHABA&tb- m=isch&iact=rc&uact=3&dur=353&page=9&start=161&ndsp=22&ved=0COoBEK0DMEw4ZA

%2F23andme-doing-free-genetic-tes%2F&h=327&w=515&tb-25 FDA: Warning Letter Online: http://www.fda.gov/iceci/enforcementactions/warningletters/2013/

ucm376296.htm, [accessed on: 6.8.2014].