Metzler meets Fraunhofer
Does data make us healthy? How big data is changing the healthcare market
Data is a valuable resource in the 21st century. In the past, data was collected for a specific purpose. Big data goes beyond that: huge amounts of data are generated and stored. With the aid of self-learning computer programs, different databases can be linked and searched for patterns of behaviour and correlations.
The opportunities, risks and possible side-effects of big data in the healthcare sector are enormous. Professor Stefan Wrobel, Director of the Fraunhofer Institute for Intelligent Analysis and Information Systems (IAIS), and Dr. Johannes Reich, personally liable partner of Metzler Bank, who is responsible for Metzler’s Corporate Finance and IT divisions, discussed big data in the healthcare market at the joint “Metzler meets Fraunhofer” discussion platform in June 2016.
Can you give a concise definition of big data?
Wrobel: Big data refers to a comprehensive perspective on the changes arising from the fact that today we are confronted with an increasing amount and range of data on all aspects of the real world. Big data is the digital twin of the physical world and it does not simply mean the technical aspects but also refers to the resulting changes for companies, markets and business models.
Reich: For me, big data is a collective term for phenomena which have one thing in common: the creation of a new meta-reality that directs actions on the basis of the simultaneous and interlinked mass production, analysis and synthesis of real-time data of all types.
Wrobel: Big data constitutes a profound change in how we view the world, which is today far more driven by data than in the past when it was based on models.
Do people have access to information about where data on them is stored, by whom, and for how long?
Wrobel: Fortunately, it is already possible to grant access to those who are interested in obtaining information on stored data in very many areas. Examples include creditworthiness evaluations, and some internet companies offer now such possibilities. In principle, wherever data is collected these days, a signed declaration of consent setting out the type and extent of the data to be collected and its intended use is required. However, in future it will be desirable to give individuals the freedom of choice. Anyone who wants to use intelligent services or other specific data-based services should be able to do so, while those who do not want to receive personalised services should be able to ensure that data on them are not stored.
Reich: I’m somewhat sceptical about that, and also quite self-critical as far as my own behaviour goes. Therefore, I do not believe it is realistic to assume that the average consumer and the average citizen will obtain meaningful information on what types of data and meta-data on him or her exists where, why and in what form. Part of the problem is that the theoretical insight of the individual is likely to be limited by what is practicable and by the ability to exploit customary everyday practice. At the same time, there is a blatant problem with how individuals recognise or fail to recognise algorithmic intelligence: who is aware of what they do not know?
Big data analyses do not provide any explanation of the relationships they reveal. In your view, what insights are gained from the purely statistical relationships shown by big data?
Wrobel: This is a very important question because big data will increasingly confront us all with statistical relationships in the form of observed correlations. It is very important to realise that these do not necessarily reflect causal connections, but are simply statistically secured correlations. Even so, that represents a big increase in our knowledge because often the existence of a statistical correlation was not known in the past. As a surprising new fact, it can therefore trigger further examination and clarification.
Reich: Knowledge is power! And that includes awareness of correlations! Even assumed or alleged knowledge is power if it is turned into a usable form of reality through conditioning - or perhaps we could say belief. People are eager for explanations: and it is irrelevant whether the interpretation of reality or causal connections comes from reading tea leaves, throwing chicken bones or the innards of sacrificial animals. Statistical relationships or correlations have at least an equally high socio-psychological claim to provide an explanation, in fact sometimes they can be better than shamanistic methods.
Wrobel: In many areas of application the mere awareness of a correlation may in itself be a sufficiently interesting and intelligent service, while in others it will be necessary to examine the underlying causes to ensure that the basis for action is absolutely reliable.
Reich: In any case, it is vital to understand that correlations can create causal connections. It must not necessarily be the other way round. If our global society is not made aware of this fact and the associated consequences, further development of artificial intelligence will become an existential threat to the individual and to entire social structures.
Which sectors are pioneers in the use of big data and the development of opportunities to evaluate data?
Reich: The pioneers are definitely the IT and software development sectors, along with the advertising industry and various forms of e-commerce, securities trading and weather forecasting.
Wrobel: A historical view shows that extensive data sets were used in the retail sector from a very early stage. Many classic examples of the use of analytics and big data come from this field, in other words, optimising the offering of retail stores and naturally also the internet. The financial community has also been familiar with predictive processes for a long time. But now big data is used relatively widely in all sectors.
Reich: And big data processes are already comparatively well developed in areas where very large volumes of structured data have to be generated or collected and processed immediately in real time.
Wrobel: A study we did for the Federal Ministry for Economics Affairs and Energy has clearly shown that the applications for big data and its actual use already exist across all corporate functions and sectors.However, that does not mean that all companies are equally advanced in this respect; there are some significant differences between sectors.
Where do you see the greatest potential for big data in the healthcare sector?
Wrobel: In the use of big data for the healthcare sector, the initial focus is naturally on people as patients.
Reich: With a great deal of optimism, I see the highest potential for personalised and thus highly targeted approaches to the prevention and treatment of chronic diseases.
Wrobel: In clinical and medical research the increased availability of data is already opening up new approaches to developing active ingredients and combating disease.
Reich: In particular, the development of highly effective new drugs with low side effects could benefit from the use of big data.
Wrobel: Taking a broader view, the personal health sensors that are already available have great potential for future use, for example, by improving access to medical care or by supporting a healthy lifestyle. Such approaches are particularly interesting where providing regular medical care is difficult. We are already seeing that smartphone-assisted services can make a meaningful contribution in the absence of doctors. However, big data is naturally also bringing changes to healthcare markets - from personalised insurance offers which create a new mechanism to optimise medical care. Such things have the potential to check the threat of further cost increases.
Are there areas of the healthcare market where big data applications are already established?
Reich: Presumably the most advanced are research and development processes for medical active ingredients and medicines, especially in genetically based fundamental research.
Wrobel: General analytical methods have been used successfully in active ingredient research for many years to replace expensive laboratory experiments with computer-based experiments and forecasts. In many cases, modern active ingredient research would be inconceivable without such assistance. Companies are even considering raising the value of data through suitable cooperations. With regard to public research we can certainly say had it not been for the collaboration based on big data we would not have been able to make such advances, for example in research into fundamental cellular mechanisms.
Reich: There are also far-reaching approaches in diagnostics. The concept behind them was actually developed decades ago under the term “expert systems” but they did not become established due to inadequate computer power. However, non-complex cost-benefit calculations suggest that we could be about to see that the diagnostic capability of outstandingly well-trained medical staff could be superseded by the diagnostic capacity of machines based on big data.
Healthcare data is sensitive. How do you rate the risks of misuse?
Wrobel: Indeed, healthcare data is among the most sensitive data that can be collected about people. Special protection of such data is therefore necessary.
Reich: Nearly a decade ago, Juli Zeh examined the loss of individual sovereignty over personal data and especially health-related data vividly and highly plausibly in her novel and play “The Method”. The self-determination, freedom, integrity and health of individuals is directly linked to the ability of the individual to defend the sovereignty of their personal data and the disclosure of such data to the collective or an authorised representative of the collective in the future. However, in order to defend themselves, individuals have to be able to recognise and name the risk existing in and as a collective. That is something we should be increasingly sceptical about.
Wrobel: The Fraunhofer organization has a substantial record of investment in research related to realising secure infrastructures, especially in the field of healthcare. We have often drawn attention to the fact that further action is needed. At the same time, we are looking at the possibility of privacy preserving analytics in many fields that use big data. In a nutshell, for many applications, a specific knowledge of the individual is not necessary. Instead, it is often possible to use data abstraction, where it is technically possible to guarantee that the individual is no longer identifiable. Based on the principle of data sovereignty, we are in favour of building a secure data room in which companies and consumers can exchange data on the basis of certified principles and agreed rules. For this purpose, we have invented the “Industrial Data Space”. This is not intended solely for industry: we are currently applying the principles to medical information.
The insurance sector is considering offering behaviour-related health insurance tariffs. Is that a move that makes sense?
Wrobel: Insurance is based on the principle that a community of insured people jointly bear the individual risk so that the consequences are manageable for the individual. However, for a long time now it has been customary for insurance tariffs to be based on the risk of the insured group of people. In the healthcare market, in the long term we therefore need to find a fine line between adjusting premiums to risk and the principle of solidarity on which insurance is based.
Reich: That is a tempting and dangerous direction, the further we move along it.
Prof. Dr. Stefan Wrobel is Professor of Computer Science at University of Bonn and Director of the Fraunhofer Institute for Intelligent Analysis and Information Systems IAIS. He studied computer science and artificial intelligence in Bonn and Atlanta, Georgia/USA (M.S., Georgia Institute of Technology) and obtained his PhD at the University of Dortmund. After holding positions in Berlin and Sankt Augustin he was appointed Professor of Practical Computer Science at Magdeburg University, before taking up his current position in 2002. In addition, since 2014 he is one of the directors of the Bonn-Aachen International Center for Information Technology (b-it). Professor Wrobel’s work is focused on questions of the digital revolution, in particular intelligent algorithms and systems for the large-scale analysis of data and the influence of Big Data/Smart Data on the use of information in companies and society. He is the author of a large number of publications on data mining and machine learning, is on the Editorial Board of several leading academic journals in his field, and is an elected founding member of the “International Machine Learning Society”. As Speaker of the “Fraunhofer Big Data Alliance”, vice-speaker of the “Fraunhofer Information and Communication Technology Group“ and speaker of the “Fachgruppe Knowledge Discovery, Data Mining und Maschinelles Lernen”, a Special Interest Group of the German Computer Science Society, he is engaged nationally and internationally in pushing forward the benefits of digitization.
Dr. Johannes Reich is personally liable partner of the bank B. Metzler seel. Sohn & Co. KGaA and a member of the Partners’ Committee B. Metzler seel. Sohn & Co. Holding AG. He is responsible for the business area Corporate Finance and for the departments Corporate Communications, Legal and Information Technology. He started his banking career with M.M. Warburg & Co. in Hamburg. Later he worked for Morgan Stanley in London and Frankfurt/Main. Dr. Reich graduated from Karlsruhe University with a master’s degree in business administration and engineering and received his doctorate from Bamberg University. He worked as a scientific assistant at Bamberg University and RWTH Aachen University.