Artificial Intelligence in Swedish Business and Society
Artificial Intelligence in Swedish Business and Society
Artificial Intelligence in Swedish Business and Society
VR 2018:09
summary
Title: Artificial intelligence in Swedish business and society - Analysis of development and potential. Summary
Author: Vinnova
Series: Vinnova Report VR 2018:09 (Summary in English of Vinnova Rapport VR 2018:08 ”Artificiell intelligens i svenskt näringsliv och samhälle”)
ISBN: 978-91-87537-76-9
ISSN: 1650-3104
Published: May 2018
Publisher: Vinnova - Sveriges innovationsmyndighet/Sweden´s innovation agency
Registration no: 2017-05616
Cover picture: Getty Images
Vinnova’s vision is for Sweden to become a leading global player in research and
innovation, and a country that is attractive for investment and entrepreneurship.
Each year, Vinnova invests around SEK 3 billion in fostering innovation. Most of these
funds are allocated via calls for proposals in which companies, public sector actors and
other organisations apply for funding. All investments are continuously monitored and
evaluated and we regularly analyse the impacts of our investments.
The Vinnova Report series includes evaluations, analyzes and reports about our
programmes, funded projects or assignments we’ve given others.
Preface ______________________________________________________________________________ 5
Introduction _________________________________________________________________________ 7
3
4
Preface
• identifying and analysing opportunities inherent in the use of AI and adjacent technologies in Swedish
industry and in Swedish business and the public sector generally,
• identifying and describing Sweden’s current position within the AI area, in particular in relation to
available demand for and supply of skills through education and training, but also in relation to
research and public initiatives to apply AI in various industries and sectors of society,
• analysing where Sweden stands in terms of taking opportunities inherent in AI and highlighting
bottlenecks that may be limiting factors in relation to utilisation.
The final report must be delivered to the Government Offices of Sweden (Ministry of Enterprise
and Innovation) by 30 April 2018. Vinnova submitted an interim report to the Government
Offices of Sweden (Ministry of Enterprise and Innovation) on 12 February 2018.
Göran Marklund
Deputy Director General
Head of Operational Development
5
6
Introduction
A central part of the analysis is to create an understanding of driving forces, opportunities, obstacles and links
between significant factors for AI-based value creation in business and the public sector.
The issues dealt with by the assignment are wide-ranging, and many different aspects of the
development of business and society are significant in relation to these issues. Accordingly, this
analysis cannot claim to be comprehensive. This summary describes Vinnova’s assessments and
conclusions of the studies, data and expert opinions on which the analysis is based.
7
Sweden’s AI capability
Applications of AI have already been of great importance for the development of internet
platforms, information retrieval, image recognition and automated translation, but the
practical impact of AI has been limited in large parts of business and in public activities in
Sweden. However, during the last decade, access to data in electronic form and computing
power has increased very quickly, which has considerably improved the conditions for AI
applications in various activities.
Sweden’s value creation potential in the use of AI within business and public services is great. Most
assessments identify a growth potential that is twice as fast with large AI utilisation in the economy
compared with a low AI utilisation.
The potential for improved quality and efficiency in the public sector is great. Additionally,
there is very considerable potential in developing and implementing AI solutions for
environmental and social challenges in society. Accordingly, artificial intelligence can
contribute to Sweden’s possibilities of achieving the goals in the 2030 Agenda for Sustainable
Development.
A considerable increase of AI applications in business, the public sector and society has not only
a potential to improve the quality and efficiency in various operations, as well as increased
growth and improved welfare; AI developments will also generate new challenges through the
development and adjustment processes that will become necessary. The following challenges
will become important as a consequence of increased AI utilisation:
8
Innovation leadership in all sectors and at all levels will become very significant for the AI
development in Sweden. The dynamics in the labour market will increase considerably as the AI
applications in business and public operations become more frequent. Adaptability of
individuals and operations will become increasingly important. This will place very high
demands on leadership skills for business development and the ability to support individuals in
readjustment and upgrading of skills. Therefore, the momentum, skills and other requirements
for such adaptability must be significantly strengthened.
The net effects of labour dynamics for the economy are largely very uncertain. Based on historical development
and new scenarios, there is no reason to assume, however, that the creation of new jobs will be slower overall
than the pace of the jobs that will disappear.
Regulatory developments and the rules regarding data and data access will be crucial for the
development of AI. Such rules must strike a balance between the fundamental needs for
privacy, ethics, trust and social protection and data access necessary for the development of
value-creating AI applications. This requires that drivers and competences to participate in
innovation processes are significantly strengthened among public authorities and experts in
charge of regulatory and regulatory monitoring. One important aspect is that such public
authorities and experts should cooperate directly with other stakeholders in the R&D and
innovation processes where new AI applications are developed.
Leadership and governance for a safe and value-creating transformation of society as a whole
must be strengthened considerably. Knowledge about how increasing AI usage may affect the
development of society and what measures may contribute to minimizing the risks of adverse
effects of AI are deemed highly undeveloped. Competence development in relation to social
aspects of AI therefore needs to be strengthened. The ability to prepare system analyses that
create understanding of how different drivers, factors and processes affect each other are very
significant in this context. However, analyses that form the bases for various policy areas are
often conducted with a systems perspective that is too narrow, linked to specific areas of policy
and based on narrow questions. Often, they are also based on a method repertoire for necessary
systems analyses that is too narrow. While initiatives are taken to promote the utilization of AI,
research, analytical capacity and processes for system analyses need to be developed
significantly.
Important areas
Artificial intelligence will be important for Sweden’s future competitiveness and innovative
strength in all sectors and industry branches, that will all be affected by the development of AI.
Therefore, it is difficult to identify areas of application for AI where Sweden is particularly well-
placed. However, the following broad and mutually dependent areas of application for AI are
expected to be particularly important for the development of both Swedish business and
society.
• Industrial development - development of products and services and manufacturing and service
processes
• Travel and transports - autonomous vehicles, logistics and transport infrastructure
9
• Sustainable and smart cities - transport systems, energy and waste, education and healthcare systems
• Health - products, services and processes for diagnostics, drugs, and healthcare
• Financial services - service development in finance, insurance and payment systems
• Security - defence, civil contingency, police and customs
The opportunities for AI applications are different in different areas. Many different factors will
be important in the development and these will be strongly mutually dependent on each other.
The following factors and the interaction between these will be important in all activities:
• Business and operational models - for some companies and public operations the value creation
potential of AI is apparent, while others cannot yet perceive a benefit as clearly.
• Drivers - for some companies AI is already a significant competitive factor, while others still lack clear
drivers, and the drivers are generally weak in the public sector.
• Data access - in most areas, lack of data access is a crucial limit on the development of business and
operational models are based on AI applications.
• Competence - limited AI skills in companies and public activities, among both managers and
employees, hampers the development of AI in most operations.
Business and operational models, data access and competence are mutually dependent and
therefore heavily affected by each other in companies and public operations. Without clear
prospects of business benefits, the drivers of AI-based investments are inhibited. If the business
benefit is not clear, AI competence is also not viewed as an important factor for value creation
and efficiency, affecting recruitment patterns and competence development. Limited AI
competence, at managerial level and among employees, makes it difficult, in turn, to develop AI
based business and operational models. Data access and possibilities of combining different
data will be fundamentally significant for purposes of identifying the applications that can be
developed. Data restrictions that inhibit or eliminate the development of AI-based products and
processes weaken the drivers for AI investments, Figure 1.
10
Figure 1 Mutual dependencies among important factors for the development of AI applications
Important prerequisites for a positive development of the interaction between the above factors
will include:
• Innovation leadership for the development of AI applications and the ability to lead business
restructuring and to support individuals in readjustment and upgrading of skills.
• Labour market models that give individuals drivers and favourable terms for labour dynamics and life-
long learning lay the foundation for continuous adjustment in the labour market.
• Data regulations for many AI applications restrict the development considerably. Such regulations often
have strong links to privacy, ethics, trust and ownership rights.
• Social solutions for digital security, integrity, ethics, trust and safety, striking a balance between the
fundamental needs for data access for AI development and social development needs.
• Critical mass and international attractiveness in environments for research, education and innovation
that are characterized by efficient cooperation among different functions and market participants.
• Collaboration among companies, public operations, research institutes, universities and university
colleges will be crucial in realizing Sweden’s AI potential.
11
Increased AI usage will also place significantly higher demands on individuals to continuously
change tasks and continuously renew their competence. This in turn will require development
and adaptation of drivers and social security systems for transitions in working life, presenting
new challenges in terms of labour market policies and for the participants in the labour market.
Successful AI development in different areas will depend on both specialist and broad
competence within AI and domain and organizational competence for different AI applications.
Access to data is closely linked with regulatory conditions relating to data and data
management. Data regarding individuals and individual behaviour is of crucial significance for
many AI applications. Data access is determined by how companies and public operations
develop and safeguard data in their operations and collaboration, and by the development of
regulatory conditions for data generation and data access. Therefore, society’s regulatory and
ethical management of privacy issues, data security and title to data will be very important in
the context of AI development.
Internationally strong environments for research, education, and innovation will therefore be
important for Sweden’s innovative strength and international attractiveness for leading AI
competence and corporate AI development. Cooperation to achieve a critical mass in such
environments will be crucial. Cooperation will also be important to link regulatory
development and labour market development with innovation processes for AI applications. It
is difficult to identify any area that is as dependent on cooperation between different
stakeholders and across sectoral borders as AI. A positive AI development in Sweden thus
demands efficient cooperation among many different actors and functions in society.
Sweden’s AI capability
There are many different factors that are significant in relation to Sweden’s ability to develop
and use AI. Different factors also play roles of different magnitude within different sectors and
different branches of industry. Table 1 is an overall analysis of Sweden’s Strengths, Weaknesses,
Opportunities and Threats (SWOT).
12
Table 1 Overall SWOT-analysis of Sweden’s AI capability
STRENGTHS WEAKNESSES
Technology-friendly population AI competence hard to recruit
High level of technological skills Lack of competence for digital business models
Qualified researchers and engineers Universities and university colleges have weak drivers for
Good domain competence regarding processes flexible professional training
Good data access Many SME have limited resources and competence
Slimmed-down organisations hamper competence
Excellent IT networks development
Infrastructure for data traffic
Many digitalised processes IT infrastructure not always accessible and stable
High level of automation IT maturity varies within value chains
Lack of coordinated security initiatives
Major international technology-driving companies Automation systems are often based on old technology
Efficient public operations
Good innovation ability Unclear ownership and data rules
Ability to solve complex problems Uncertainty concerning future data access regulation
Long experience in security-critical solutions Lack of AI standards
Difficult to cross-check data
Culture of cooperation Failures in data quality and structure
Developed innovation system
Efficient value chains and ecosystems Fragmented municipal sector
Ability to build consortia Poor cooperation among county councils
Lack of state control
Few and unfocused R&D initiatives in AI
OPPORTUNITIES THREATS
Increased innovation pace with AI Regulatory development does not interact with the AI
Use access to large volumes of data requirements
Use our quality register for AI analysis Privacy issues are dealt with differently in Sweden
Intellectual property rights could be threatened
New technical opportunities via system connections
New functions and improved quality of products and Poor AI competence at management level
services Poor AI competence weakens development momentum
Increased efficiency in production and processes Regional differences in AI competence
System potential in new value chain connections AI competence leaving Sweden
New work methods and new organisation methods Fears and unrealistic expectations
New interesting and attractive jobs Long implementation timelines inhibit investment
Improved work environment Lack of AI investments inhibits competitiveness
Sweden could be a test bed for AI development The rest of the world invests more and faster than
Sweden has a high attractiveness level internationally Sweden
Develop cooperation around AI-development Sweden will not become a test bed for new AI solutions
Develop cooperation around AI-implementation Large dependence on systems suppliers
Use the cooperation capacity of the research institutes Increasing vulnerability in systems
Poor IT security increases social risks
Train existing AI competence
Develop regulations promoting data access Simple jobs are disappearing and unemployment is rising
Develop policies promoting system development Trust and confidence in the future is inhibited by the
development of AI
Distrust as AI failed in the 80s and 90s
13
AI competence will be of crucial significance in order to realize Sweden’s AI capability. While
computer science skills are important, access to software engineers will be crucial, since
successful AI development often requires extensive software development. Access to AI
competence will be a significant challenge, since there is a global lack of such expertise. AI
applications are expected to increase drastically in the next few decades. Global demand for AI
competence will therefore increase sharply, which means that an already significant lack of AI
competence is expected to grow further.
The ability of universities and university colleges in Sweden to adapt the focus of their research
and education to the rapid and multidisciplinary changes that AI generates is poor. There are
many indications that further education must account for most of the education system’s
adaptation in catering to the AI requirements of business and society. However, universities
and university colleges find it difficult to develop and operate brief independent courses in close
interaction with and directly adapted to working life, i.e. for life-long learning in general and AI
in particular. The forms of cooperation between universities and university colleges with
business and society need to be developed, especially to achieve a faster and improved
adaptation of the educational offer to existing needs.
The digitisation of the educational institutions needs to be developed, both at a basic level and
in universities and university colleges. This applies both to the use of digital technology and AI
in courses as well as the utilization of digitalisation opportunities for management and
administration. It is likely that successful education and research institutions will, in the
future, have high digitisation rates in both core and support activities. AI will be significant in
this context.
Bibliometric data indicate that Swedish AI research has, overall, limited international
competitiveness. It is a generally accepted view that development within AI, both in research
and commercially, is dominated by the USA, with China being the main contender, while
Europe has tended to lose ground, relatively speaking. An analysis of the contributions to the 19
highest ranked AI conferences since 2010 strongly supports this view. American researchers
participate in nearly half of all conference contributions. Researchers from China have the
strongest increase and their share is nearly one fifth. The presence of Swedish researchers at
the same conferences is very modest, with only 0.6 percent of all contributions in 2014-2017
and a downward trend compared to previous years. The participation per capita at the
14
conferences is many times greater for Singapore, Switzerland, and Israel than for Sweden and
significantly larger also for Australia, Canada, Finland, and Denmark.
The situation changes considerably if these comparisons are widened to include all the articles
published in journals classified as belonging to AI as a subdivision of computer science - mainly
by strengthening China’s position. The average citation rate for articles with authors from
China was indeed just a third of the citation rate for articles with authors from the United
States a few years ago. Nevertheless, there were almost as many articles with Chinese writers
among the 10 percent most cited as with American writers. For Sweden the wider picture is
somewhat more favourable than that for highly ranked conferences. This applies in particular
to the average citation rate that gives a positive picture of the quality level of AI research in
Sweden. However, when comparing the number of publications among the 10 percent most
cited, the distance to Singapore and Switzerland is great.
The great attention that has surrounded artificial intelligence globally in recent years has its
background in an increased use of machine learning in various applications. So-called deep
learning has yielded particularly striking results. The increased use of AI is reflected only to a
small extent in conferences and journals with AI specialists as primary target audience. An
analysis based on key words with a close connection to deep learning gives the impression that
Sweden, from a relatively favourable starting point at the turn of the century when technology
was still in its infancy, failed to join in the growing experimentation with deep learning that
has taken place in some other countries and successively lost ground. Fortunately, there is
evidence that a positive shift in this trend occurred in 2017.
Patent data is another indicator of Swedish development capability. Since around 2010, AI is a
rapidly growing patent area, and today it is one of the largest patent areas globally. AI-related
patenting is strongly dominated by major IT companies such as Samsung, Microsoft, IBM and
Google, but as the use of AI is widened to new areas, the group of companies applying for AI-
related patents is also widening.
Based on data on patent applications to the patent authorities in the USA, Europe and Japan,
Sweden’s development capability in AI appears neither particularly weak, nor particularly
strong. Sweden is ranked 13th internationally in AI patenting and just under 1 percent of all the
world's AI patents in recent years have involved inventors from Sweden. In a closer comparison
with five other smaller countries, Sweden’s AI patenting per capita is a little ahead of Canada,
equal with Switzerland and Denmark, a little behind Finland and significantly behind Israel.
In Sweden, as in the other five countries above, the number of AI patented companies has
increased significantly in recent years. However, Ericsson strongly dominates Swedish AI
patenting with computer networks and the operation of mobile communications networks as
special strengths, which also resulted in these areas being the strengths of Sweden, a position
that has also been strengthened. Smart transports and vehicles is one of the fastest growing
sub-areas for patenting within AI and of particular importance for Sweden with several major
vehicle companies in the country. However, in this area of patenting Sweden has been unable
to keep up with the development in the rest of the world and seen its share of patenting halved.
15
Artificial intelligence is still a relatively unestablished research area in the Swedish research
system. Within undeveloped research areas, there is often a need to invest with special
initiatives in building research expertise and research environments before a strong demand
for such skills emerges from business or society. In Sweden, it is difficult for academic
institutions to develop internationally strong research environments within AI through their
own strategic initiatives.
16
Concerted effort to boost Sweden’s AI
development
The goal of a Swedish AI strategy should be to turn Sweden into a leading country for the development and
application of AI for sustainable growth and social development.
Since the correlations and dependencies are strong between different factors that are
significant for AI development, it is important that a national policy stimulates cooperation
between all significant stakeholders, so that a mutually reinforcing development momentum is
generated.
Sweden’s greatest opportunities for competitiveness within AI lies within a mutual interaction between
innovative AI application in business and innovative organization of society.
This requires a well-developed strategic cooperation among participants in business, the public
sector, research and education. Accordingly, not only strategy as such is important, but through
the strategy process itself, the necessary knowledge, competencies, leadership and collaborative
relationships can be strengthened among the various actors involved in the process.
A national strategy based on a broad and inclusive process should be designed and include most areas of policy
and public authorities.
Strategic areas
In a national concerted effort to achieve Sweden's AI potential, it is important to purposefully
prioritize the development of each of the following areas so that a mutually reinforcing
interaction between them is generated:
• Drivers for companies and public operations to streamline and develop new value-creating solutions
based on artificial intelligence.
• Cooperation in research, development, data access and competence development for AI innovation,
linking requirements within different value chains and sectors for a combined development momentum.
• Further education and basic education for continuous supply of AI skills in the labour force, which
requires a renewal of the education system.
• Research investments in AI for excellence and internationally leading research and innovation
environments for advanced AI research and R & D collaboration.
• Regulatory development for data access, data integration and data ownership promoting AI innovation
and safeguarding integrity, ethics and data security.
17
• Infrastructure development promoting research, development and testing of AI applications and AI
security, integrity and rules in connections between various data, value chains and societal areas.
• Labour market development promoting the labour force mobility required by AI, which places
significant demands on transition leadership in companies, the public sector and politics.
Figure 2 Policy areas and mutual dependencies in a concerted effort for Sweden’s AI-potential
Prioritised initiatives
In the context of a national concerted effort, some government initiatives are particularly
important for positive development. These should be initiated before a national strategy is
available, while they should also be central parts in such a strategy.
Further education in AI, which is well adapted to the labour market, should be stimulated with special
initiatives for a rapid development of such education and processes for these.
Without special policy initiatives, this development will not take place at the pace required to
realise Sweden’s AI potential. The development of further education in AI and dimensioning of
these courses needs to take place in a close dialogue between academic institutions, companies
and the public sector.
Internationally leading collaborative environments with a critical mass in research, education and innovation,
with advanced data and technology infrastructures, need to be developed.
Without special policy initiatives for this purpose, the necessary concerted effort will not take
place. This in turn inhibits the innovation processes for AI-based value creation. As of 2012, the
18
Swedish Foundation for Strategic Research (Stiftelsen för Strategisk Forskning) (SFF) has made major
investments in cutting edge research within AI. The Swedish Knowledge Foundation (Stiftelsen
för kunskaps- och kompetensutveckling) (KK Foundation) has also made several significant
investments in research and education within AI at Swedish academic institutions and new
universities. Wallenberg AI, Autonomous Systems and Software Program (WASP), started in
2015, is a concerted effort for research and education. From and including 2017 a strongly
reinforced 10-year investment in AI in the context of this venture will be made. In the
cooperative programs financed by Vinnova in collaboration with the Swedish Energy Agency
and the Swedish Research Council Formas, there has been a drastic increase of AI projects in
recent years. AI has also increased drastically in the innovation projects for small and medium
size enterprises (SME) financed by Vinnova. This development is significant because a major
part of the challenges in both research and education concern the use of AI methods in real
applications.
Access to data for both education and research as well as application-focused development projects in AI
should be an important priority in a national concerted effort.
Society’s regulatory and ethical management of privacy issues, data security and data
ownership will be very important in the context of AI development. To enable several parties to
access the same data, in many cases special agreements governing the use of such data will be
required.
Test beds for AI development based on access to important data should be an important focus for a critical
mass and efficient cooperation in research and innovation environments.
The role of public organizations in making data available is important for AI applications in the
public sector, business and societal development at large. Companies should also be encouraged
to open their data to promote cooperation and attractive research and innovation
environments in Sweden.
19
The development of basic education and research-level education in AI for width, excellence and cooperation
with industry and the public sector should be prioritized.
In addition to further education, the following parts should be included in the development of
higher education for stronger AI competence:
• Investment in industrial doctoral students and industrial postdoctoral students for testing AI solutions.
• Appointment of new combined teaching and research positions:
with recruitment in Sweden and internationally,
of which part with shared positions between institutions, academic institutions and companies.
• Commitment to education by companies, institutions and the public sector, with practical AI
experience.
• The educational system should provide basic computer science for students in many areas.
Drivers and conditions for academic institutions to digitize courses, research training and support operations
need to be reinforced.
The Governance and Resource Investigation's Committee (Styr- och resursutredningen, Strut)
on how governance and resources for universities should be developed will be important for the
development of drivers and conditions for academic institutions. The final report of the
Investigation will be presented in December 2018.
Drivers and conditions for individuals to take more responsibility for their own competence development needs
to be strengthened to encourage active life-long learning and job mobility.
Such drivers and conditions will be important to stimulate and facilitate individuals' capacity
for adaptation and lifelong learning, which will become increasingly important for the future's
skills supply. This, in turn, will be of major significance for achieving a labour market dynamics
and human trust in the rapid technological renewal and social transformation. In this context,
network-based courses, such as MOOCs, which are not time- or location-dependent, should be
covered by the study funding system. In addition, tax breaks, incentive programs or subsidies
should be considered to create opportunities for individuals to take greater responsibility for
their own skills development.
20
Government control and drivers for an innovative public administration will be very important for a value
creating AI development that strikes a balance between business innovation and privacy, ethics and digital
security.
For many public operations, the drivers to develop and apply AI are weak. These drivers
originate in the governance of state and municipal operations. How such governance
stimulates and requires innovation management and adaptation ability will be of great
importance for AI development in Sweden. Trust-based governance, investigated by
Tillitsdelegationen (the Swedish Trust Delegation) is an important part of such developed
governance.
Developed state governance will be crucial for the utilisation of AI ‘s transformative potential in the
development of systems solutions to address important social challenges and to achieve the targets in the
2030 Agenda for Sustainable Development.
The drivers of the public sector to develop new system solutions that address social challenges,
where AI can play a central role, are even weaker than for an AI-based development in the
context of individual public operations. Powerful development that addresses social challenges
requires developed administrative governance, creating clear drivers for an innovative and
collaborative public administration. Such a developed administrative governance needs to
generate innovative drivers that cross policy areas, government boundaries, administrative
boundaries and geographical boundaries.
21
Examples of AI projects
22
Improve healthcare with AI and image analysis
AIDA is a project promoting research and innovation within artificial intelligence and medical image analysis, with
the goal of improving healthcare.
There are great opportunities to improve image examinations using AI. For example, analysis of large data volumes
can create a higher accuracy in diagnostics.
We know through research that current AI technology is very powerful, but because it is not adapted to healthcare, it
has not yet achieved any notable benefit in this field. In AIDA, academics, healthcare and industry meet to create AI
innovations. This cooperation results in decision support that creates patient benefit in healthcare. Cooperation
between humans and machines is in focus. By using these respective strengths optimally, healthcare will
significantly improve.
The project is led by Linköping University and today includes seven hospitals, five academic institutions, one major company and four
small companies.
23
Deep learning improves the process industry
The Swedish process industry produces huge amounts of data and there is an untapped potential in these data,
which can be utilized with new tools. The project ”Deep Process Learning” uses data from a board machine at
BillerudKorsnäs in Gävle to investigate how analysis of large volumes of data with the help of deep learning
algorithms can improve quality and efficiency.
By analyzing large volumes of data, so-called deep learning, there is potential for increasing the competitiveness of
the Swedish process industry through increased productivity, quality and flexibility.
Deep learning has the human brain as a model, and with the aid of algorithms, a system builds knowledge gradually,
just as humans do. When the system has built knowledge, it can perceive patterns and better control our
processes. In the longer term, the project hopes to inspire other industries to use their large volumes of data for
development and process optimization. In 2018, the parties in the project will start a new initiative to develop web-
based courses to spread knowledge on how these technologies can be used.
The project is a cooperation between BillerudKorsnäs, PulpEye, Peltarion, FindIT and RISE SICS Västerås, who leads the project.
24
AI improves breast cancer screening
Today's breast cancer screening has reduced mortality, yet about 1,500 women die annually in Sweden of breast
cancer. AI could help reduce the number of deaths due to breast cancer by detecting tumors earlier.
By training deep learning with over one million mammography images combined with clinical data from the Breast
Cancer Registry, decision support is produced in the "AI for Breast Cancer Screening" project.
The goal of the decision support is to identify easily assessed mammographs and to identify the women who benefit
most from a supplementary study. The digital decision support can easily be distributed across the country and
reduce regional differences in the screening system. The hope is that this will provide socioeconomic value with safer
diagnosis and improve the efficiency of the mammography process in Swedish healthcare.
Karolinska University Hospital coordinates the project AI for Breast Cancer Screening with the participation of among others Regionalt
Cancercentrum Stockholm-Gotland and Sectra AB.
25
26
Vinnova’s publications May 2018
05 Staden som arena för innovation - En studie av 02 From green forest to green commodity chemicals -
transformativ kapacitet, kommunens roll och Vinnovas påverkan Evaluating the potential for large-scale production in Sweden for
three value chains
06 Digitalisering - mer än teknik. Kartläggning av svensk
forskning och näringslivets behov 03 Innovationstävlingar i Sverige - insikter och lärdomar
07 Årsbok 2017 - Svenskt deltagande i europeiska program för 04 Future Smart Industry - perspektiv på industriomvandling
forskning och innovation 05 Det handlar om förändring - Tio år som följeforskare i Triple
08 Artificiell intelligens i svenskt näringsliv och samhälle Steelix (For English version see VR 2016:05)
- Analys av utveckling och potential. Slutrapport (For English 06 Evaluation of the Programme Multidisciplinary BIO
summary see VR 2018:09) - The strategic Japanese-Swedish cooperation programme 2005 -
09 Artificial intelligence in Swedish business and society 2014
- Analysis of development and potential. Summary (For full 07 Nätverksstyrning av transportinnovation
Swedish version see VR 2018:08) 08 Ersättningssystem för innovation i vård och
VR 2017: omsorg – En studie av åtta projekt som utvecklar nya
01 Att skapa förutsättningar för innovation - Erfarenheter ersättningsmodeller
från ”Idéslussar i kommuner - förstudie 2015” VR 2014:
02 Testbäddar inom hälso- och sjukvård och äldreomsorg - 01 Vägar till välfärdsinnovation - Hur ersättningsmodeller och
Portföljutvärdering av Vinnovas program impact bonds kan stimulera nytänkande och innovation i offentlig
03 Samband mellan immateriella tillgångar, innovation verksamhet
och ekonomisk tillväxt - Två kunskapsöversikter 02 Jämställdhet på köpet? - Marknadsfeminism, innovation och
04 På jakt bland forskare och managementkonsulter - normkritik
Klinisk forskning och praktiknära kunskapsutveckling inom 03 Googlemodellen - Företagsledning för kontinuerlig innovation i
managementområdet en föränderlig värld
05 Utvärdering strategiska innovationsprogram – Första 04 Öppna data 2014 - Nulägesanalys
utvärderingen av Innovair, BioInnovation, IoT Sverige, Smartare 05 Institute Excellence Centres - IEC -En utvärdering av
Elektroniksystem, SIO Grafen och Swelife programmet
06 Why manufacture in Sweden? Strengths and best practice 06 The many Faces of Implementation
- a summary of ”Flaggskeppsfabriken” (For full version in Swedish
07 Slututvärdering Innovationsslussar inom hälso- och
see VR 2016:07)
sjukvården
VR 2016:
01 Third Evaluation of VINN Excellence Centres - AFC,
BiMaC Innovation, BIOMATCELL, CESC, CHASE, ECO2, Faste, Vinnova Information
FUNMAT, GHz, HELIX, Hero-m, iPack, Mobile Life, ProNova, VI 2018:
SAMOT, SuMo & WINGQUIST 01 Årsredovisning 2017
02 Third Evaluation of Berzelii Centres - Exselent, UPSC & 03 Delredovisning av handlingsplan för
Uppsala Berzelii jämställdhetsintegrering på Vinnova - 2015-2018
03 NOVA - Verktyg och metoder för normkreativ VI 2017:
innovation (for English version see VR 2016:06) 01 Forskning inom gruv- och mineralområdet - En studie av
04 Forskning och utveckling för ökad jämställdhet styrkor och samverkan
- Följeforskning om Vinnovas regeringsuppdrag avseende 02 Projektkatalog 2016 Utmaningsdriven innovation Steg
behovsmotiverad forskning för ökad jämställdhet 2013-2015 1 - Initieringsprojekt
05 This is about Change - Ten years as an on-going evaluator of 03 Projektkatalog 2016 Utmaningsdriven innovation Steg
the Triple Steelix initiative (For Swedish version see VR 2015:05) 2 - Samverkansprojekt
06 NOVA – tools and methods for norm-creative 04 Projektkatalog 2016 Utmaningsdriven innovation Steg
innovation (for Swedish version see VR 2016:03) 3 - Följdinvesteringsprojekt
07 Flaggskeppsfabriken - Styrkor i svensk produktion (For 05 Årsredovisning 2016
summary in English see VR 2017:06)
06 Challenge-Driven Innovation - Societal challenges as
08 Flaggskeppsmetodiken - En arbetsmetod för industriellt opportunities for growth (for Swedish version see VI 2016:07)
erfarenhetsutbyte
07 FFI Årsrapport 2016 - Samverkan för stark svensk
fordonsindustri och miljöanpassade samt säkra transporte
VI 2016: 02 Samverkansuppgiften i ett historiskt och
01 Projektkatalog Utmaningsdriven innovation Steg institutionellt perspektiv
1-2015 - Initieringsprojekt 03 Långsiktig utveckling av svenska lärosätens
02 Projektkatalog Utmaningsdriven innovation Steg samverkan med det omgivande samhället - Effekter av
2-2015 - Samverkansprojekt forsknings- och innovationsfinansiärers insatser
03 Projektkatalog Utmaningsdriven innovation Steg 04 Företag i Tåg- och järnvägsbranschen i Sverige - 2007-
3-2015 - Följdinvesteringsprojekt 2013
04 Årsredovisning 2015 05 FoU-program för Små och Medelstora Företag -
05 FFI Årsrapport 2015 - Samverkan för stark svensk Metodologiskt ramverk för effektanalyser
fordonsindustri och miljöanpassade samt säkra transporter 06 Small and beautiful - The ICT success of Finland & Sweden
06 Innovation för ett attraktivare Sverige - 07 National Research and Innovation Councils as an
Sammanfattning Instrument of Innovation Governance - Characteristics
07 Utmaningsdriven innovation - Samhällsutmaningar som and challenges
tillväxtmöjligheter (for English version see VI 2017:06) 08 Kartläggning och behovsinventering av test- &
08 Vinnväxt - A programme renewing and moving Sweden ahead demonstrationsinfrastruktur
VI 2015: VA 2014:
01 Insatser för innovationer inomHälsa 01 Resultat från 18 VINN Excellence Center redovisade
2012 - Sammanställning av enkätresultaten. (For English version
02 FFI Årsrapport 2014 - Samverkan för stark svensk see VA 2014:02)
fordonsindustri och miljöanpassade samt säkra transporter
02 Results from 18 VINN Excellence Centres reported in
03 Social innovation - Exempel 2012 - Compilation of the survey results. (For Swedish version see
04 Social innovation VA 2014:01)
05 Årsredovisning 2014 03 Global trends with local effects - The Swedish Life Science
06 Sweden needs FFI (for Swedish version see VI 2015:10) Industry 1998-2012
07 Innovation för ett attraktivare Sverige - Underlag till 04 Årsbok 2013 - Svenskt deltagande i europeiska program för
regeringens politik för forskning, innovation och högre utbildning forskning och innovation.
2017-2020 - Huvudrapport 05 Innovations and new technology - what is the role of
08 Förutsättningar för innovationspolitik i Sverige - research? Implications for public policy. (For Swedish version see
Underlag till regeringens politik för forskning, innovation och VA 2013:13)
högre utbildning 2017-2027 - Analysrapport 06 Hälsoekonomisk effektanalys - av forskning inom
09 Replaced by VI 2016:07 programmet Innovationer för framtidens hälsa.
10 Sverige behöver FFI (for English version see VI 2015:06) 07 Sino-Swedish Eco-Innovation Collaboration - Towards a
new pathway for shared green growth opportunity.
11 Replaced by VI 2017:06
08 Företag inom svensk massa- och pappersindustri - 2007-
2012
Vinnova Analysis 09 Universitets och högskolors samverkansmönster och
VA 2017: dess effekter
01 The Energy Industry in Sweden continues to grow -
Analysis of companies in the energy industry 2007-2014 - business
segments, age structures, gender equality and competence. (For
Swedish version see VA 2016:05)
02 Hårdare, lättare och snabbare - Effektanalys av ett urval av
Vinnovas materialrelaterade satsningar
03 Årsbok 2016 - Svenskt deltagande i europeiska program för
forskning och innovation
04 The automotive industry in Sweden - A cluster study
VA 2016:
01 Vinnväxt - Ett innovativt program i takt med tiden
02 Årsbok 2015 - Svenskt deltagande i europeiska program för
forskning & innovation
03 Effektanalys av Vinnväxt-programmet - Analys av
effekter och nytta
04 Chemical Industry Companies in Sweden - Update
including data for competence analysis
05 Energibranschen i Sverige fortsätter växa - Analys
av företag i energibranschen 2007-204 - branschdelar,
åldersstrukturer, jämställdhet och kompetens. (For English version
see VA 2017:01)
06 Omvandling och fasta tillstånd – Materialvetenskapens
etablering vid svenska universitet
07 Svensk konsultsektor i ny belysning -
Utvecklingstrender och dynamik
VA 2015:
01 Årsbok 2014 - Svenskt deltagande i europeiska program för
forskning & innovation
Production & layout: Vinnova communication
May 2018
Vinnova is helping to strengthen Sweden´s innovation capacity