BIG DATA – BIG IMPACT PROGRAMME

BIG DATA TECHNOLOGIES – Current use and future potentials University November 13, 2014

BIG DATA – BIG IMPACT CONFERENCE 3

WELCOME

It is a great pleasure to welcome you to the conference on Big Data – Big Impact.

A close collaboration between businesses, We warmly welcome and value your input governmental parties and researchers plays a today. This conference aims at gathering people crucial role in the continuing development of with various backgrounds with the intention that solutions that may help solve grand challenges you this afternoon will leave Aarhus University of our society, both nationally and globally. Big with new contacts that can help you make great data holds a huge potential in helping us solving ideas through the use of big data. some of these challenges. We are delighted to welcome you at what prom- The focus of this day is the practical use of big ises to be a highly educational and inspirational data. Through four keynote speeches and several meeting on ‘Big Data – Big Impact’. case presentations it is our hope that you will find inspiration for future solutions. To fulfil this inspira- tion there is need for professional collaboration, and there is a large potential for a close collabo- ration between your company or institution and Yours sincerely, Aarhus University; a potential we hope to inspire Allan Flyvbjerg engagement in. Dean

BIG DATA – BIG IMPACT CONFERENCE 5

PROGRAMME

08:00 Registration

09:15 Welcome Allan Flyvbjerg, Professor, Dean of Faculty of Health, Aarhus University M3 Auditorium

09:30 Big Data – Possibilities and Challenges Lars Arge, Professor, Dept. of Computer Science, Aarhus University M3 Auditorium

10:15 Coffee break in The Hall

10:30 IBM Healthcare: Developing Watson to Impact a Complex Industry Patrick McNeillie, Chief Physician Researcher, Watson Healthcare, IBM M3 Auditorium

11:15 Big Data – Big Security Problems? Ivan Damgård, Professor, Dept. of Computer Science, Aarhus University M3 Auditorium

12:00 Lunch in The Hall

13:00 Case presentations The afternoon case presentations will be divided in 4 parallel thematic sessions: • Societal Processes • Business • Healthcare • Technology The Hall or M3 Auditorium

16:00 Big data as a growth factor in Danish Business (in Danish) Sigrid Wilbeck, Head of Section, Danish Business Authority M3 Auditorium

16:45 Closing note Mogens Nielsen, Professor, Dept. of Computer Science, Aarhus University M3 Auditorium BIG DATA – BIG IMPACT 6 CONFERENCE

CASE PROGRAMME

SESSION: SOCIETAL PROCESSES (The Hall)

Steered by Jan Ifversen Vice Dean, Faculty of Arts

13:00 - 13:20 Opening Up Municipal Data Torben Glock, City of Aarhus & Martin Brynskov, AU

13:20 - 13:40 Discussions

13:45 - 14:05 Measuring impact across social media Anne Jensen, NorthSide Festival & Anja Bechmann, AU

14:05 - 14:25 Discussions

14:30 - 14:50 Teledialogue Between Social Workers and Children Placed in Foster Care or at Institutions Stinne H. Mathiasen, Municipality of Herning & Lars B. Andersen, AU

14:50 - 15:10 Discussions

15:15 - 15:35 Big News: Exploring Newspaper Archives and How People Use Them Jens F. Berg, Visiolink &Tonny S. Jensen, State and University Library

15:35 - 15:50 Discussions

Further information on page 16-10 BIG DATA – BIG IMPACT CONFERENCE 7

CASE PROGRAMME

SESSION: HEALTHCARE (The Hall)

Steered by Michael Hasenkam Vice Dean, Faculty of Health

13:00 - 13:20 skolesundhed.dk – An interactive web based child health platform Thomas Jensen, Opus Consult & Carsten Obel, AU

13:20 - 13:40 Discussions

13:45 - 14:05 Patient-Reported Outcome Data in Primary Health Care Philipp Harbig, Lægerne & Kaj S. Christensen, AU

14:05 - 14:25 Discussions

14:30 - 14:50 Safe Delivery App Anders Nejsum, Visikon & Christina Braüner, AU

14:50 - 15:10 Discussions

15:15 - 15:35 Danish Hip Arthoplastry Register (DHR) – P of the Danish Orthopedic Common Database Torben B. Hansen Central Region & Alma B. Pedersen, AU

15:35 - 15:50 Discussions

Further information on page 20-23 BIG DATA – BIG IMPACT 8 CONFERENCE

CASE PROGRAMME

SESSION: BUSINESS (The Hall)

Steered by Per Baltzer Overgaard Vice Dean, Faculty of Business & Social Sciences

13:00 - 13:20 Development of Service Logistics based on Position and Task Data from Super Hospitals Mikkel Harbo, Femming B. Thomsen, SYSTEMATIC & Kaj Grønbæk, AU

13:20 - 13:40 Discussions

13:45 - 14:05 Waste Transportation Logistics Inge Werther, DAKOFA & Sanne Wøhlk, AU

14:05 - 14:25 Discussions

14:30 - 14:50 Using Secure Multiparty Computation to Facilitate Secure Auctions and Confidential Benchmarking Jacob Pagter, ALEXANDRA INSTITUTE, Kurt Nielsen, PARTISIA & Ivan Damgård, AU

14:50 - 15:10 Discussions

15:15 - 15:35 Increasing Customer Profitability Through Predictive Modelling of Likely Consumer Purchasing Patterns Christian S. Maltesen, COOP Danmark & Morten H. J. Fenger, AU

15:35 - 15:50 Discussions

Further information on page 24-27 BIG DATA – BIG IMPACT CONFERENCE 9

CASE PROGRAMME

SESSION: TECHNOLOGY #1 (M3 Auditorium)

Steered by Kurt Nielsen Vice Dean, Faculty of Science & Technology

13:00 - 13:20 Cleaning Ocean Floor Scanning Data Freek van Walderveen, SCALGO, Ole Kristensen, EIVA & Lars Arge, AU

13:20 - 13:40 Discussions

13:45 - 14:05 Air Quality at Your Street Christian L. Fogh, Environmental Protection Agency Denmark & Matthias Ketzel, AU

14:05 - 14:25 Discussions

14:30 - 14:50 Inferring and Analyzing Transportation Behavior based on Mobile Sensing Data Rolf Sode-Carlsen, RAMBØLL & Kaj Grønbæk, AU

14:50 - 15:10 Discussions

15:15 - 15:35 Flood Risk Screening Morten Revsbæk, SCALGO, Jeppe S. Jensen, COWI & Lars Arge, AU

15:35 - 15:50 Discussions

Further information on page 28-31 FOTO BIG DATA – BIG IMPACT CONFERENCE 11

CASE PROGRAMME

SESSION: TECHNOLOGY #2 (The Hall)

Steered by Mogens Nielsen Vice Dean, Faculty of Science & Technology

13:00 - 13:20 Human Sensory Perception Based Healthy Eating Leslie Jørgensen, Rynkeby Foods & Derek V. Byrne, AU

13:20 - 13:40 Discussions

13:45 - 14:05 Big Data in Farm Animal Genetics Anders Fogh, Knowledge Center For Agriculture & Bernt Guldbrandtsen, AU

14:05 - 14:25 Discussions

Further information on page 32-33 BIG DATA – BIG IMPACT 12 CONFERENCE

LARS. ARGE PROFESSOR AARHUS UNIVERSITY

BIG DATA – POSSIBILITIES AND CHALLENGES

The pervasive use of computers, as well as Arge is a Professor of Computer Science at tremendous advances in the ability to ac- Aarhus University and Director of the Danish quire, store and process data, has resulted in National Research Foundation (Danmarks a spectacular increase in the amount of data Grundforskningsfond), and Center for Massive being collected. There are many exciting Data Algorithmics (MADALGO). MADALGO possibilities for use of the data in both science pursues a broad basic research agenda with- and industry – big data can potentially have a in efficient algorithms for big data, but also very big impact! However, there are obvi- works with industry partners and researchers ously also many challenges in unlocking this from other fields on using algorithms research potential. Often its not clear what information advances in practical applications. Thus Arge can be extracted from the data and even less has for example both obtained significant clear how to extract it. theoretical and practical results in relation to big terrain data. Arge is an elected member In his talk, Arge will give examples of big data of the Royal Danish Academy of Sciences and its impact, especially highlighting how and Letters and The Danish Academy of university research has resulted in innovative Technical Sciences, a Fellow of the Associ- applications and solutions through collabora- ation of Computing Machinery (ACM), the tion between researchers and industry. He will recipient of the Danish Minister of Research describe some of the computer science re- Elite Research Award, and a co-founder of search being performed at Center for Massive the company SCALGO that markets software Data Algorithmics at Aarhus University, along and services in connection with big terrain with its applications in relation to for example data processing flood risk estimation. BIG DATA – BIG IMPACT CONFERENCE 13

PATRICK MCNEILLIE CHIEF PHYSICIAN RESEARCHER WATSON HEALTHCARE, IBM

IBM HEALTHCARE: DEVELOPING WATSON TO IMPACT A COMPLEX INDUSTRY

IBM’s approach to healthcare focuses on sultant. He attended the University of North improving physician-patient interactions, Carolina at Chapel Hill School of Medicine, mainly the collection, analysis, and com- earning his doctorate in medicine. He has munication of healthcare information. With been active in basic science and clinical on- the assistance of the medical community cology research for more than 10 years, with IBM has initially developed technologies to more than 30 peer-viewed publications. Prior assist medical research, clinical practice, and to joining IBM he spoke extensively about healthcare management. innovations within cancer treatment. In 2012 he won the Pillsbury Award for Outstanding In this talk Dr McNeillie will discuss the data Oral Presentation in Clinical Medicine. In problem with healthcare and how IBM Wat- 2013 he took a year off of residency training son is being developed to provide a solution. and joined IBM as a Post-Doctoral Research Physicians face two big data problems, Fellow. During this year he work along-side medical knowledge gained from research IBM computer scientists in applying the and patient information documented in cognitive technologies of IBM Watson to records. IBM has partnered with several of the healthcare. He decided to extend his time top academic medical centers in the world at IBM and was promoted to Chief Physi- to develop Watson to more effectively bring cian Researcher of the Medical Sieve Grand medical knowledge to physicians, to help Challenge. The goal of this project was summarize patient records, and --combining to develop technology for medical image these tasks—to assist physicians in making the analytics. Recently, he transitioned from IBM most appropriate decisions. Research to Watson Transformation to assist in bringing information technologies to the Dr Patrick McNeillie is currently working with healthcare industry. IBM Watson Healthcare as an Oncology Con- BIG DATA – BIG IMPACT 14 CONFERENCE

IVAN DAMGÅRD PROFESSOR AARHUS UNIVERSITY

BIG DATA – BIG SECURITY PROBLEMS?

Companies, public organisations and intelli- Ivan Damgård is a professor of Computer gence services collect huge amounts of data Science at Aarhus University and director of at an ever increasing speed. This holds great the Center for research in the Foundations promise for the services we can expect from of Electronic Markets (CFEM), supported computer systems but also raises questions by the Danish Strategic Research Council. from a security and privacy point of view. Damgård’s research is focused on crypto- In the talk, Damgård will shed light on some logic methods for protection of data, more of the stories told about the subject and try to specifically he has obtained a number of distinguish myth from reality. For instance: Is fundamental results on secure distributed there always a good reason to collect data? computing. Damgård is a fellow of the Inter- Are our identities and personal data safe on national Association for Cryptologic Research the Internet? Do we have any privacy left out (IACR) and the Editor in Chief of the Journal there? Should we care? If yes, what can we of Cryptology. He is also a co-founder of the do about it? companies Cryptomathic and Partisia. BIG DATA – BIG IMPACT CONFERENCE 15

SIGRID WILBECK HEAD OF SECTION DANISH BUSINESS AUTHORITY

BIG DATA AS A GROWTH FACTOR IN DANISH BUSINESS

Much has been said about the potential of Sigrid Wilbeck works in the field of big data Big Data, however, what impact does it have and has worked for the national growth on the Danish businesses? A year ago The team, referred to as “Digital Growth and Danish Business Authority decided to analyze ICT” and subsequently worked on a plan for the impact of Big Data in Danish businesses, growth in this area, to be published this Fall and the results from a survey conducted by 2014. Currently Sigrid Wilbeck is developing the IRIS Group was published in the report the use of a new public data base, the Virk “Big data as a growth factor in the Danish Data - a data base run by The Danish Busi- business”. The report gives a picture of which ness Authority which serve Danish compa- types of companies, that are the primary us- nies who will start using public data in their ers of Big Data in Denmark, and highlights the businesses. various kind of barriers the companies face when dealing with big data. BIG DATA – BIG IMPACT 16 CONFERENCE

OPENING UP MUNICIPAL DATA

SESSION: SOCIETAL PROCESSES Torben Glock Martin Brynskov Project Manager Associate professor Citizen Services Interaction technologies City of Aarhus Aarhus University

The Municipalities and the Public Sector in gen- and the democratic value by opening up Data eral in Denmark, contains process and generate in one Municipality, this project has the chance immense amounts of data. Much of this data to lead the way for other Public Institutions to resides in separate legacy systems and is not gradually opening up their Data Vaults. easily accessible for analysis, use and inspection. The reason is sometimes technical, sometimes By opening up the Public Sector data, the private legal, sometimes ethical and often grounded in System Developers has a chance for a better a Public Sector cultural discourse where Open rooted empirical based knowledge, thereby Data is not the norm. creating higher quality solutions to ensure amore SMART and efficient Public Sector. Working with Aarhus University, the City of Aarhus has over the years begun to turn this mul- An alternate approach to the cooperation of the titude of challenges into a resource for planning Public-,the Private- and the Civil society within a and producing the conditions for better lives project context of Open Data. between the systems, together with its citizens. The project aims at supporting the Open Data paradigm in a context of the Public Sector. By investigating the opportunities, the risks, the utility BIG DATA – BIG IMPACT CONFERENCE 17

MEASURING IMPACT ACROSS SOCIAL MEDIA

SESSION: SOCIETAL PROCESSES Anne Jensen Anja Bechmann Partner and online manager Associate Professor NorthSide Festival Head of Digital Footprints Research Group, Aarhus University

Social media plays an increasingly important streams from the NorthSide festival 2014. In role in cultural events worldwide. At concerts and collaboration with NorthSide festival and with festivals you often see participants with smart- permission from the Danish Data Agency the phones in the air to document the event. These group also carried out qualitative interviews and digital footprints of the participants create a large data retrieval from all personal data streams. digital social layer that connects to the event. Based on mainly public streams this presentation The digital footprint that is left behind during a will highlight some of the patterns that occur festival therefore is important in at least two ways. in the data collected. Rethink Impact funds the In a media research perspective studying social research project. media footprints in pre-planned events may add to existing research methodologically because a The study of social media footprints allows for new software tool and methodological consider- methodological development and research ations are needed in order to study the activities into the relationship between events and digital across platforms. Also the footprints tell us what (social media) layers. characterizes the digital social layer of the festival. Which social media platform do they use, when The project seeks to include a business partner to do participants upload content, what kind of con- make precise accounts for the methodological tent, with whom do they share this content, and do challenges in using the software and approach- they tag themselves into the public data streams? es in real life event settings.

In a business strategic perspective finding pat- terns in these social media footprints may help the event managers to optimize resources and communication in the social media department. The Danish music festival NorthSide has social media as one of their strategic focus areas. The Digital Footprints Research Group has built software to collect all public social media data BIG DATA – BIG IMPACT 18 CONFERENCE

TELEDIALOGUE BETWEEN SOCIAL WORKERS AND CHILDREN PLACED IN FOSTER CARE OR AT INSTITUTIONS

SESSION: SOCIETAL PROCESSES Stinne Højer Mathiasen Lars Bo Andersen Udviklingskonsulent, phd. post doc. Municipality of Herning Aarhus University

Danish children placed in foster care or at On this backdrop, the research project Teledia- institutions are under the formal custody of logue seeks to develop improved and continuous social services departments. Accordingly, these communication between social workers and departments collect a wide array of data on children through chat and video conferencing the children which they manage through case software. management software. In this way, children not only become more There are many different data providers sup- intimately acquainted with the social worker plying the social services departments with holding custody but also become more active information on the children. Foster parents, contributors to the complex set of information biological parents and pedagogues regularly through which they are represented at the social call or write the social worker in custody with services departments. various concerns. Likewise, grades are collected from schools, psychologists write capability as- Teledialogue allows for interdisciplinary and ex- sessments and teachers comment on academic plorative research into the relationship between abilities. citizens and state by coupling social problems with technical ones, and vice versa. All these inputs are journalized in the depart- The project also seeks to include stakeholders in ment’s database which enable changing social developing a specific concept for an improved workers to supervise the welfare and upbringing and IT-enabled communication between of the children on an informed foundation. children and social workers - which includes chil- dren’s active participation in the data collection However, the children themselves do not partake through which the social services departments directly in this continuous data collection. Rather, operate. their dealings with social workers is often limited to biannual meetings and, in addition, mediated by pedagogues or other adults. BIG DATA – BIG IMPACT CONFERENCE 19

BIG NEWS: EXPLORING NEWSPAPER ARCHIVES AND HOW PEOPLE USE THEM

SESSION: SOCIETAL PROCESSES Jens Funder Berg Tonny Skovgård Jensen CEO Director – National Library Division Visiolink State and University Library

Newspaper archives are an important part of a as the archives become an important source for society’s historical heritage by offering the public research. and paying newspaper readers attractive con- tent from a historic past. Historic content has become available to the public from the State and University Library as The historic archives cover a number of news- well as from newspaper websites owned by the paper’s entire history. Usually digital archives media companies. contain newspapers from the last up to 200 years and are fully accessible from web, tablets and As a public/private partnership the project pro- smartphones. vides access to historic media content from both libraries and media portals, which strengthens New methods to access historic content means the overall approach to use historic sources. not only lots of new content to explore and analyze, but also new levels of data collection, We expect researchers to use the archive for and when millions of pages are available on traditional academic work such as search, col- searchable platforms, the collection of important lection of sources, reading and analyzing – just knowledge on usage and behavior is available. faster and easier, making it possible to cover more ground. But we also expect to attract re- By exploring the usage of 32 million digitized searchers in the emerging field of digital humani- newspaper pages The State and University ties, exploring new types of research questions Library provides a valuable source to data on by using data mining, working in academic fields usage and behavior of library users and newspa- like language, cultural studies and social studies. per readers. Through a strategic public/private partnership between the The State and University Library and the Danish media companies, mil- lions of people get access to millions of pages. Through investigating server logs and event data there’ll be access to relevant knowledge on the interest of big events or special occasions as well BIG DATA – BIG IMPACT 20 CONFERENCE

SKOLESUNDHED.DK – AN INTERACTIVE WEB BASED CHILD HEALTH PLATFORM

SESSION: HEALTHCARE Thomas Jensen Carsten Obel CEO Professor Opus Consult Research Program for Mental Child Health Aarhus University

In Denmark we have unique epidemiological for Mental Child Health, department of pub- resources to develop an understanding of what lic health, Aarhus University (www.iupgrowth. determines health in a life-course perspective. au.dk), the participating Danish municipalities The current project adds to the unique combina- (especially Odense and Copenhagen) and Opus tion of large register based cohorts and ad hoc Consult, as Danish software developer company cohorts by providing an interactive platform to specialized in clinical databases. The develop- share data between professionals and research- ment of skolesundhed.dk started in 2007 and at ers. The core facility is epidemiological feedback. this point 20-25 of the 98 Danish municipalities Professionals (health nurses, psychologists, social use the application. The development has been workers, doctors) are provided feedback at an financed by Trygfonden and the participating individual level (clinical decision support) and municipalities. at group level (comparison with other school classes, centers). On a general population level In the next phase of this project the Committee all children at 1st, 5th, 7th and 9th grade provide of Health Information (www.sundhedsformidling. detailed information on somatic and mental dk) will take over the responsibility of implement- health for the health nurse at an individual level. ing the application at a national level. One of the This information is used to focus the clinical core elements of the application is the Strength meeting. For children with need for social, and Difficulties Questionnaire (SDQ), a measure learning support or with mental health problems of mental wellbeing which is the worldwide most specialized modules have been developed to extensively used. The inventor, Robert Goodman, match the professional most relevant infor- will be engaged in the further development of mation. Other professionals such as teachers, the project providing a platform for international administrators and psychologists have access extension. At this point the priority will be trialing to stratified and thus anonymous data only. The out in the Nordic countries. At AU level this major research potential lies in analytic as well application is the core element in our collabora- as experimental epidemiology by combining tions with department of Computer Science, the these data with population registers. Alexandra institute, department of economics and DPU (teaching faculty). Skolesundhed.dk has been developed in collaboration between the Research Program BIG DATA – BIG IMPACT CONFERENCE 21

PATIENT-REPORTED OUTCOME DATA IN PRIMARY HEALTH CARE

SESSION: HEALTHCARE Philipp Harbig Kaj Sparle Christensen General Practitioner Associate Professor Lægerne Vennelystparken Mental Health in Primary Care (MEPRICA) Aarhus University

Web-based data entry software can be used to has so far been developed for diagnosis and obtain accurate and detailed data on patient- monitoring of 1) depression, 2) anxiety disorders reported outcome (PRO) in the health-care and 3) blood pressure measurements at home. system. Such data can be used for diagnosis, Sundhedsmappen.dk will soon form part of the monitoring of treatment effects and research scheme developed by the Danish Quality Unit for purposes. PRO data in primary care have General Practice to promote better data quality traditionally not been systematically collected, in primary care. The system will, therefore, be analysed and archived. AU researchers have implemented nationwide. PRO digitalization will initiated development of a web-based system provide research with important and solid data, for entering PRO measures and transmitting data which will allow e.g. measurement of effects of to patient medical records in primary health psychotherapy and pharmacological treatment care. The project is funded by the Danish Ministry of depression and anxiety disorders. of Health and the Lundbeck Foundation. The software, which is launched under the name ‘Sundhedsmappen.dk’, is currently applied in 110 primary care clinics in Denmark.

The new data entry system builds on collabo- ration between the research network Mental Health in Primary Care (MEPRICA) at Aarhus University and medical software consultants from Opus Consult in Risskov. PRO software BIG DATA – BIG IMPACT 22 CONFERENCE

SAFE DELIVERY APP

SESSION: HEALTHCARE Anders Nejsum Christina Braüner CEO Midwife, MPH and Ph.D. Visikon Aarhus University

Maternity Foundation works for a world where The Safe Delivery APP is part of both the Mater- childbirth means life, not death. The causes of nity Foundation and Visikon’s work to position maternal and newborn death and disability itself as strong actors within the emerging field of are complex, but we know that if women have mHealth. mHealth (the use of mobile phones for access to family planning, are giving birth with health promotion) is a relatively new field with skilled attendance and can access emergency an abundant unexplored potential to strengthen obstetric and newborn care in case of complica- health interventions in both rural areas as well tions then the probability of maternal and new- as the western world. In this project, research is born survival is high. Maternity Foundation works being conducted at many levels. Both local in to promote these conditions so that all women Ghana with questioners and interviews - and in can give birth safely – also in low-income coun- Danmark with data pull of user interactions. You tries. We have an integrated focus on improving will be presented with the tale of how a local the quality of and access to health services. This need inspired to a solution and the research pro- includes empowering and mobilizing the local ject now been conducted by Christina Bräuner. community to be in a better position to safeguard The emphasis in the presentation will be on data their reproductive health and train health care collection, endpoints and how all the stakehold- workers and hereby build up the clinical capacity ers, including the local Danish business, can gain of the health care system. from this collaboration and sharing of data.

In addition to conventional approaches Materni- ty Foundation also pursues new and innovative methods in order to improve mothers’ and new- borns’ heath and survival. The use of animated educational material presented via a mobile app is one of those initiatives. The solution is de- veloped based on preliminary research done by Danish researchers and clinicians in close corpo- ration with animators form the company Visikon. The solution “Safe Delivery APP” are being tested in countries in Africa. BIG DATA – BIG IMPACT 23 conference

DANISH HIP ARTHOPLASTRY REGISTER (DHR) – PART OF THE DANISH ORTHOPEDIC COMMON DATABASE

SESSION: HEALTHCARE Torben Bæk Hansen Alma Becic Pedersen Professor MD, PhD Executive consultant Staff specialist, Clinical associate professor KEA Central Denmark Region Aarhus University Hospital

In Denmark we do close to 10.000 primary total DHR is working in cooperation with the other hip arthroplasties (THR) per year. The Danish Hip Nordic arthroplasty registers (Norway, Finland Arthroplasty Registry (DHR) was initiated by the and Sweden) in the Nordic Arthroplasty Register Danish Orthopaedic Sociaty on January 1, 1995. Association (NARA). The overall aim is to identify DHR is a nationwide clinical database on pri- patients, surgeon and surgery related factors that mary THR, revisions, and follow-up examinations can predict worse outcome after THR, change in Denmark. There are now reported 120,988 the treatment and clinical practice and improve primary and 19,282 revisions to the DHR in the patient care. NARA collaboration has resulted in period 1995-2012. All orthopaedic departments several publications allowing studies on patients in Denmark report to the register. The registered groups which are too small to be studies in single data include preoperative, peroperative, and countries. DHR is also a member of the Interna- postoperative information, prospectively collect- tional Society of Arthroplasty Registries (ISAR). ed using a standardized forms. Registration of DHR is an example of the work done by The primary operations and revisions is compulsory, Department of Clinical Epidemiology, Aarhus whereas recording of follow-up examinations is University - and research done base on registry voluntary. However, patients can be followed-up data for which Denmark is renowned. The data- several years after total hip arthroplasty, de- base is not directly connected to the Electronic pending on the hospital´s routine only. Through Health Records or The National Patient Registry. annual report data, participating orthopaedic However, using civil registration number, the departments are able to compare their perfor- data in the DHR have been be linked to several mance in order to improve quality of treatment. Danish administrative and clinical databases The Database continually incorporates new for research and clinical monitoring purposes, elements to accommodate the clinical needs. including Danish National Registry of Patients, Latest the focus on issues concerning the use of Danish Transfusion Database, Danish National Metal on Metal (MoM) prostheses have led to an Prescription Database and Civil Registration expansion of the DHR. System. The potential in the use of inductive statistics in connection to the DHR, and the many The DHR have a major impact on decision mak- databases like it, is considered to be of interest to ing on many levels from the political, economi- researchers from different faculties than Health at cal to the clinical, made in the operation room. AU, private businesses and clinicians. BIG DATA – BIG IMPACT 24 CONFERENCE

DEVELOPMENT OF SERVICE LOGISTICS BASED ON POSITION AND TASK DATA FROM SUPER HOSPITALS

SESSION: BUSINESS Mikkel Harbo Flemming Bent Thomsen Kaj Grønbæk Project Manager Vice President Professor Systematic Engineering A/S Systematic Engineering A/S Dept. of Computer Science, Aarhus University

In the PosLogistics projects we have developed 1) unsupervised route and travel time estima- a system for optimizing indoor route and task tions based on as-traveled-data from planning for service work primarily undertaken smartphones by orderlies in Hospitals. 2) smart logistics based on indoor positioning and learning from historical data Service worker Phones deliver position and 3) Automatic task phase recognition and sensor data to a backend that analyses indoor context-aware feedback on wearable travelling and task performance to make unsu- devices. pervised route and task planning algorithms that optimizes planning of service tasks. The presentation will feature technical solutions, Temporary or permanent changes to the work analysis, and experiences from real world de- environment, e.g. broken elevators, corridors ployments will be presented. closed for painting, are datamined automatically from the collected phone data and seamlessly used to change future route and task planning. The system utilizes WiFi-based indoor positioning and it has huge potentials for reducing waiting times and optimizing services at super-hospitals.

There are several research contributions from this project: BIG DATA – BIG IMPACT CONFERENCE 25

WASTE TRANSPORTATION LOGISTICS

SESSION 4: BUSINESS Inge Werther Sanne Wøhlk Ressource and Waste Specialist Associate Professor DAKOFA CORAL Aarhus University

The increased focus on sustainability has different waste collection systems as regards significantly impacted the waste management transportation. This analysis can be used as business in Denmark and abroad. Increasing the input in the strategic decisions regarding system amount of recycling has become a focal point design. for public authorities. With the Resource Strategy 2013, the Danish Ministry of the Environment has formulated a strategic goal of increasing To be able to estimate the cost of a waste the percentage of recycling and reuse and of system, a major component lies in the estimation decreasing the amount of waste in general. of the mileage needed for the waste collection Recycling and recycling costs have thus become and transportation vehicles. We use a quanti- issues of vital importance for both society and tative research approach based on Operations environment. Research to find near optimal solutions to these large transportation problems and use these to The counties in Denmark have a high degree of evaluate the relative cost of the systems. freedom regarding their waste strategies. Many different systems are being considered for collec- tion and handling of waste and recyclables, both regarding the degree of sorting at the house- holds and regarding the technologies used. We analyze different systems and combinations of systems in order to find the relative costs of BIG DATA – BIG IMPACT 26 CONFERENCE

INCREASING CUSTOMER PROFITABILITY THROUGH PREDICTIVE MODELLING OF LIKELY CONSUMER PURCHASING PATTERNS

SESSION: BUSINESS Christian S. Maltesen Morten H. J. Fenger Consumer Insights Director Ph.D. student Coop Danmark A/S Aarhus University

A well-assorted supermarket may wish to sell ap- individual offerings received in general from the prox. 100,000 different products to its customers. retailer. Such a model can be predictive, i.e. based A typical consumer may wish to buy approx. 10 on what other similar consumers have purchased. products on a given day, ending up buying 20. The proposed model is not a standard market The first 10 products may be planned purchases basket analysis application, as known from Am- based on routine and supermarket advertising azon.com, since the goal is not merely to guide circulars. The other 10 products may be impulsive consumers to buy like others. Instead this model purchases chosen on the way through the super- aims at identifying latent consumer segments in market. Over time the consumer will develop her terms of the sequence in which the product port- consumption habits in this way – as a combina- folio typically is expanded. Such a model provides tion of habits adjusted for good offers and things information as to which products the consumer catching her attention on her way through the may be likely to add to her market basket in the supermarket – a potentially slow process for very future. If this addition to the product portfolio has habitual consumers. The standard means of a positive impact on the expected future value of affecting this process includes costly activities the customer to the retailer, then the retailer may of product promotions together with the costly wish to accelerate this development by providing and environmentally dubious practice of printing the customer with individual offers for products weekly colourful circulars. Many consumers opt likely to be adopted in future purchases. out of receiving these, not because they don’t want good relevant offers, but because they In cooperation with Coop Denmark A/S, trans- don’t find them being good relevant offers, or -be actional panel data for 10,000 households for cause the offers they find good are few and not all purchases over a period of 20 months are good enough to warrant the time to identify and analyzed. For the case of this presentation, it is act upon them. People are different but standard assessed how consumption of organic products marketing is one-size-fits-all. And it doesn’t. develop per consumer over time, from one prod- uct category to another. This provides insight into On the other hand, if retailers can tailor fewer of- how organic food products may be efficiently ferings to the individual consumer – offerings likely marketed using individual product offerings. The to be found interesting by this specific consumer methodology can be applied to many other – then the consumer may be more attentive to the focus areas, e.g. private labels. BIG DATA – BIG IMPACT CONFERENCE 27

USING SECURE MULTIPARTY COMPUTATION TO FACILITATE SECURE AUCTIONS AND CONFIDENTIAL BENCHMARKING

SESSION: BUSINESS Jakob Pagter Kurt Nielsen Ivan Damgård Head of Research and Innovation CEO Professor Security lab Partisia ApS CFEM, Alexandra Institute A/S Aarhus University

In many auctions and procurement scenarios, it in financial benchmarking, data mining across is essential that the bids placed by participants several databases, etc. are kept confidential. Not only would the game be unfair if a participant could fabricate a bid The first application of MPC mentioned above while knowing bids of other players, a possibly was motived by the fact that EU drastically more serious issue is that information on loosing reduced support for sugar beet production. bids reveal information on the bidder’s economic Economists from CBS and KU analyzed the situ- status that could be used against him in other ation and concluded the sugar beet production contexts. Thus it is important to have a technolo- would not survive in Denmark unless a nation- gy that can guarantee participants that their bids wide market for trading production quotas was remain confidential and that the result is correct. established – since otherwise production could Furthermore, one must be able to handle large not be moved to places where it pays off. Based nation-wide auctions where the set of bidders is on a series of basic research results, researchers potentially all players in a certain sector. While from AU research centers SIMAP and CFEM and handing all bids to a trusted third party may work software engineers from the Alexandra institute in a paper-based system, such a single point of designed and implemented a system that would attack is unacceptable in a modern digital sys- implement the required market – a so-called tem. Secure Distributed Computing (also known double auction. As a result the spin-off com- as Multiparty Computation (MPC)) offers a way to pany Partisia was formed. Partisia now runs the deliver the required security guarantees without auction regularly and now also markets auction having a single point of failure. Collaboration solutions for use in the energy sector, allowing, between AU researchers and industry has led for instance, SMEs to easily get the best offers for to the first industrial application of the technol- their energy supply. ogy, to trade EU production quotas between farmers. In a survey, 75% of the farmers said that confidentiality of their bids was important to them. Similar secure auction systems are now commercially available in the energy sector. The technology has countless other applications BIG DATA – BIG IMPACT 28 CONFERENCE

CLEANING OCEAN FLOOR SCANNING DATA

SESSION: TECHNOLOGY #1

Freek van Walderveen Ole Kristensen Lars Arge Senior developer Software manager Professor SCALGO ApS EIVA A/S MADALGO Aarhus University

Multi-beam sonar scanning equipment can be in collaboration with EIVA and Statoil, and these used to obtain very accurate and detailed – and results have then been further developed by thus very big - data about the ocean floor. It is SCALGO and integrated in general subsea soft- for example used to inspect the many pipelines ware from EIVA. The so-called S-CAN software that transport oil and gas in the ocean. Removing both result in a simpler and better cleaning unwanted data (cleaning) from scanning data process: “By means of the S-CAN feature, we has traditionally been a labor-intensive manual have reduced the time spent on cleaning by process. Collaboration between AU computer more than 25 percent, and we obtain a far better science researchers and industry has led to soft- overall result” - Per Norvald Boge, Lead Data Pro- ware for cleaning of ocean floor scanning data cessor, DOF Subsea (in EIVA product brochure). that is not only (semi-) automatic but also pro- duce better results than previously. The software is now deployed on 5 continents.

The new data cleaning software is a result of a collaboration between algorithms researchers at the basis research center MADALGO at AU, developers at MADALGO spinout company SCALGO, and subsea experts at EIVA (as well as at Statoil). Theoretically efficient algorithms have been developed (and published) by MADALGO BIG DATA – BIG IMPACT CONFERENCE 29

AIR QUALITY AT YOUR STREET

SESSION: TECHNOLOGY #1 Christian Lange Fogh Matthias Ketzel Special consultant Senior researcher Environmental Protection Agency, Denmark Aarhus University

Health impact assessment of air quality in Den- epidemiological studies to establish the link mark estimates that all air pollution from national between exposure to air pollution and health and international emission sources is due to effects. The data may also be used to study the about 3,300 premature deaths in Denmark and relationship between air quality levels and house associated with large welfare economic costs. prices or the relationship between air quality The public is concerned with air pollution and AU levels and socio-economic status improving our regularly receives requests from individuals and knowledge on environmental equity. also planners and regulators in public admin- istration about air quality in specific locations. To be able to provide air quality data for all It is therefore important to provide information locations and addresses in Denmark for the first about air quality beyond the few places where time, a number of air quality models have been air quality is measured routinely and beyond linked and all required input data for models the few locations where calculations are carried concerning emissions, meteorology, traffic data, out on an annual basis as part of surveillance of street geometry etc. have been established. air quality in Denmark. On this background an Danish Environmental Protection Agency has interactive website has been developed that been consulted in the process of development of allows the general public to view the geographic the service. DCE has financed the development variation of air quality for selected health-related based on funding provided by AU. pollutants down to any of the about 2.3 million addressed in Denmark. The service is provided free of charge. The website is expected to have many users that will look for the air quality where they live, work or their children go to school etc. It will also have an interest for people who consider to move to another place, and real estate dealers have also showed interest in the data. The air quality data also provides a basis for answering a number of research questions. For example, it provides for better air pollution BIG DATA – BIG IMPACT 30 CONFERENCE

INFERRING AND ANALYZING TRANSPORTATION BEHAVIOR BASED ON MOBILE SENSING DATA

SESSION: TECHNOLOGY #1 Rolf Sode-Carlsen Kaj Grønbæk Project Manager Professor Rambøll A/S Department. of Computer Science Aarhus University

In the EcoSense project we have developed a There are several research contributions from general big data backend to collect and analyze these projects: data from a general SmartPhone sensing app 1) collection and homogenized analysis of framework that enables advanced analysis of raw data from a diversity of smartphones traffic and transportation behavior. This gener- with different sensor characteristics. al platform is deployed for a number of green 2) Machine learning and pattern recognition in transportation projects, where SmartPhone apps collected data are used for data collection. 3) accurate transportation mode detection. 4) Robust and scalable data collection An example of a project is “Herning Bikes to the backend Moon ”, facilitated by Rambøll. It supports collec- tion of data from a large number of inhabitants The presentation will feature technical solutions, in a city, and machine learning algorithms on analysis and experiences from real world de- the server extracts the detected bike trips from ployments will be presented. the multitude of transportation trips and count individual and collectively biked kilometers as part of a biking campaign.

The different variants of Apps send data to gen- eral backend and different analysis and visualis- ations can be applied real-time or historically to the data collections. The smartphone apps have been distributed. BIG DATA – BIG IMPACT CONFERENCE 31

FLOOD RISK SCREENING

SESSION: TECHNOLOGY #1 Morten Revsbæk Jeppe Sikker Jensen Lars Arge CEO Head of section Professor SCALGO Aps COWI A/S MADALGO Aarhus University

Extreme weather events, e.g. with large amounts these results have then been implemented and of rain in a short period of time or large sea-level developed into products by SCALGO. SCAL- rise, are becoming increasingly common. These GO now markets a number of terrain software events can have catastrophic effects on housing products that are capable of handling much and other assets, for example when water gath- larger datasets than other product, along with a ers in depressions in the terrain. Thus it is impor- number of data analysis products – in particular tant to predict what areas are in risk of flooding flood risk screening products – that use big and during such events. Collaboration between AU detailed terrain models, along with other big computer science researchers and industry has datasets (such as soil types and property values), led to new flood risk screening products based which in Denmark are available as part of the on the very detailed – and thus very big - data in government free data program. In collaboration the national terrain model recently released by with COWI, one analysis product for screening of the Danish government. These products have e.g. flood risk due to extreme rain has, as mentioned, been sold to around half of the local govern- been sold to around half of the local govern- ments (kommuner) in Denmark. ments in Denmark.

The national terrain model contains more than 20 billion height points (one per 1,6x1,6 meter) and is therefore very challenging to use in various (flood risk) analyses. Standard software simply cannot handle the big data. This problem has been addresses in collaboration between algo- rithms researchers at the basis research center MADALGO at AU, developers at MADALGO spinout company SCALGO, and hydrological ex- perts at COWI. Theoretically efficient algorithms and new flood risk screening methods have been developed (and published) by MADALGO in collaboration with SCALGO and COWI, and BIG DATA – BIG IMPACT 32 CONFERENCE

HUMAN SENSORY PERCEPTION BASED HEALTHY EATING

SESSION: TECHNOLOGY #2 Leslie Jørgensen Line Holler Mielby Derek V. Byrne Product Developer Ph.D. Professor Rynkeby Foods A/S Department of Food Science, Aarhus University Aarhus University

Atempts to increase public awareness of healthy species Stevia rebaudiana. Stevia in principal is eating practices in the Nordic countries and be- the perfect replacement for sugar on a nutritional yond seem not to have led to significant changes level contributing zero calories to the diet. How- in patterns of food purchase and consumption. ever, from a human senses perspective stevia Much is understood about nutrition through is very different in perception terms to common science and the relationship of nutrition, diet sugar. Stevia depending on extract used can and health is communicated ubiquitously, but be up to 300 times sweeter and its taste has a the question remains why then are obesity and much slower onset and longer duration than that non-communicable diseases still a significant of sugar. Moreover, certain of its extracts have a problem in society as a result of detrimental bitter or licorice-like aftertaste at high concen- eating habits. trations. These sensory effects therefore present many challenges in utilizing stevia as a direct In the case we will present a Nordic perspective sweet replacer in food products. on larger data gathering which advocates a consumer, or in broader terms, a society driven From an industrial perspective food and health strategy in food perception research to develop research needs to investigate the role that inno- and propose healthy foods which can be adapt- vation in foods (e.g. sensory acceptance focused ed and tolerated from a human sensory and healthy food development aiming for formula- acceptance perspective into the diet. tion and reformulation processes with a view to enhanced adoption) and new basic research We will present evidence of practical appli- technologies (e.g. for gaining greater insight cations of sensory research in industrial food and understanding of the mechanisms under- production in Denmark and present this in the lying eating behaviour with respect to personal global health context. food identity and culture) could play in counter acting the continuing rise of food-related health The beverage industry has long revolved around problems. This focus in food and health research sugar reduction as a response to heightened if actively integrated with the input of data from calorie awareness. Thus our focus will be on the science will be effective and ensure potential for case of stevia, a calorie free sweetener and sug- a long term impact in improving societies eating ar substitute made from the leaves of the plant habits. BIG DATA – BIG IMPACT CONFERENCE 33

BIG DATA IN FARM ANIMAL GENETICS

SESSION: TECHNOLOGY #2 Derek V. Byrne Anders Fogh Bernt Guldbrandtsen Professor Team Leader Associate professor Aarhus University Nordic Cattle Genetic Evaluation Center for Quantitative Genetics and Genomics Knowledge Centre for Agriculture Aarhus University

For decades farmers and farmers’ organizations Close collaboration and mutual trust between have amassed data on the characteristics and the university and industry is key factors permit performance of farm animals. Ever increasing taking full benefit of this opportunity. amounts of data are collected in databases. They form the basis for optimizing management The synergy between university, industry and big in everyday practice on the farm. They also form data has allowed big gains in the performance the basis for selection of the best breeding ani- of agriculture. At the same time developments mals as well as research into the genetic mech- are making agriculture more sustainable by pre- anisms underlying differences in characteristics serving genetic diversity, improving breeding for through close collaboration between Aarhus animal health and improving resource efficiency University and farmers’ organizations and coops. to the benefit of the global environment. In recent years growth of data has accelerated on several fronts. Automatic collection of data Our research focus on exploiting huge datasets is introduced everywhere. At the same time to understand the genetics of farm animals and revolutionary advances in biotechnology allow optimize animal breeding to the benefit of farm- inferences on animal genetics in unprecedented ers, consumers, animals and the environment. detail. BIG DATA – BIG IMPACT 34 CONFERENCE

AU CONFERENCE STANDS 

AU CAREER DCA – Danish Centre for Food and Agriculture

AU Career is the link between students and DCA supports knowledge exchange between employers. We offer various services to students researchers at Aarhus University and authorities, to help them in their process of finding the right companies and organizations within the food student job, internship, project and full time and agricultural sector. Activities include re- position. search-based public-sector consultancy, support to national and international research cooper- Some of the services are: Career counseling, ation and strategic support to research cooper- Career Events, e.g. courses in CV and cover letter ation with companies, organizations, trade and writing, job interviews, contract negotiations etc. industry. We also offer the employers various recruitment services i.e. job postings in our AU Job- and Subscribe to our newsletter and take advantage Project bank – www.jobbank.au.dk. of our services at www.dca.au.dk

At the Faculty of Business and Social Sciences DCA – Danish Centre For Food and Agriculture and Faculty of ARTS we offer an extended Aarhus University employer branding package to employers that Blichers Allé 20, DK-8830 Tjele they can obtain by payment. Please contact us Phone: +45 8715 6000 for more information about employer branding E-mail: [email protected], and Partnership with AU Career. Web: www.dca.au.dk

As an employer or company you are welcome to contact us for more info about our career services.

Find out more: http://careercentre.au.dk and at http://phd.au.dk/career

Contact Division Manager Jette Hammer at [email protected] for further information BIG DATA – BIG IMPACT CONFERENCE 35

CONTINUING AND PROFESSIONAL DEVELOPMENT, AARHUS UNIVERSITY

Aarhus University offers a wide range of research Continuing and Professional Development at based activities within continuing and profes- Aarhus University is offered at each of the four sional development for the private and public faculties within select areas and has a spe- sector. This line of activities includes master’s and cialized administrative unit located in Herning, MBA programmes, diploma degree programmes, Emdrup and Aarhus. full-time study at part-time, customized learning programmes and different types of other training Learn more about the possibilities within continu- activities. ing and professional development at www.au.dk/evu In order to be able to cater to the needs of the corporate and public sector, continuing and professional development is organized in a specialized administrative unit responsible for the operation of the activities. RESEARCH COLLABORATION WITH AARHUS UNIVERSITY

Aarhus University has extensive collaboration with If you wish to explore the possibilities for collabora- private businesses and public authorities both tion with Aarhus University please contact our domestically and abroad. Researchers from our One Stop Service: +45 60 20 26 62 university are always open for new opportunities to collaborate on research areas of common interest. Or visit our websites: Aarhus University engages in both short term www.au.dk/videnudveksling/innovation collaborative research projects as wells as long (in Danish) term strategic partnerships. www.au.dk/en/knowexch (in English)

Aarhus University 8715 0000 Aarhus Universitet [email protected] Nordre Ringgade 1 wwwau.dk 8000