1、EIT Health is supported by the EIT, EIT Health is supported by the EIT, a body of the European Union a body of the European Union Transforming healthcare with AI The impact on the workforce and organisations March 2020 2Transforming healthcare with AI: The impact on the workforce and organisations C

2、ontents Foreword 4Foreword 4 Acknowledgements 6Acknowledgements 6 Abbreviations 7Abbreviations 7 Executive summary 9Executive summary 9 Chapter 1 Introduction 22Chapter 1 Introduction 22 1.1 AI and its potential to transform healthcare 231.1 AI and its potential to transform healthcare 23 1.2 The fo

3、cus and scope of this report 261.2 The focus and scope of this report 26 1.3 Approach and methodology 271.3 Approach and methodology 27 1.3.1 MGI analyses on the impact of automation and AI on healthcare 271.3.1 MGI analyses on the impact of automation and AI on healthcare 27 1.3.2 Expert interviews

4、 and survey 281.3.2 Expert interviews and survey 28 1.3.3 Case studies 291.3.3 Case studies 29 Chapter 2 Artificial intelligence in healthcare today 30Chapter 2 Artificial intelligence in healthcare today 30 2.1 What do we mean by AI in healthcare? 312.1 What do we mean by AI in healthcare? 31 2.2 H

5、ow recent advances have made AI in healthcare a reality 322.2 How recent advances have made AI in healthcare a reality 32 2.3 Implementation around the world 342.3 Implementation around the world 34 2.3.1 Government action 342.3.1 Government action 34 2.3.2 Private-sector investments 342.3.2 Private

6、-sector investments 34 2.3.3 A diverse research pipeline 362.3.3 A diverse research pipeline 36 2.3.4 The view from Europe 382.3.4 The view from Europe 38 2.4 Selected use cases along the AI in healthcare framework 412.4 Selected use cases along the AI in healthcare framework 41 2.4.1 Self-care, pre

7、vention and wellness 422.4.1 Self-care, prevention and wellness 42 2.4.2 Triage and diagnosis 442.4.2 Triage and diagnosis 44 2.4.3 Diagnostics 462.4.3 Diagnostics 46 2.4.4 Clinical decision support 482.4.4 Clinical decision support 48 2.4.5 Care delivery 502.4.5 Care delivery 50 2.4.6 Chronic care

8、management 542.4.6 Chronic care management 54 2.4.7 Improving population-health management 582.4.7 Improving population-health management 58 2.4.8 Improving healthcare operations 602.4.8 Improving healthcare operations 60 2.4.9 Strengthening healthcare innovation 622.4.9 Strengthening healthcare inn

9、ovation 62 2.5 From AI today to AI tomorrow 632.5 From AI today to AI tomorrow 63 Chapter 3 How will AI and automation change the healthcare workforce? 68Chapter 3 How will AI and automation change the healthcare workforce? 68 3.1 Jobs lost, gained and changed: Healthcare in 2030 693.1 Jobs lost, ga

10、ined and changed: Healthcare in 2030 69 3.1.1 Impact on employment numbers 713.1.1 Impact on employment numbers 71 3.2 How will AI and automation change the activities of healthcare practitioners? 733.2 How will AI and automation change the activities of healthcare practitioners? 73 3.2.1 Less admin

11、; more patient care 733.2.1 Less admin; more patient care 73 3.2.2 Supporting clinical activities 753.2.2 Supporting clinical activities 75 3.2.3 Easier access to more knowledge 783.2.3 Easier access to more knowledge 78 3.2.4 Patient empowerment, self-care and remote monitoring 783.2.4 Patient empo

12、werment, self-care and remote monitoring 78 3.3 New activities and new skills 793.3 New activities and new skills 79 3.3.1 A new way of interacting with patients 793.3.1 A new way of interacting with patients 79 3.3.2 Boosting digital skills in the broader healthcare workforce 803.3.2 Boosting digit

13、al skills in the broader healthcare workforce 80 3.4 Introducing new professionals in healthcare 823.4 Introducing new professionals in healthcare 82 1Transforming healthcare with AI: The impact on the workforce and organisations Chapter 4 What needs to change to encourage adoption of AI in healthca

14、re? 84Chapter 4 What needs to change to encourage adoption of AI in healthcare? 84 4.1 Quality and suitability of solutions 854.1 Quality and suitability of solutions 85 4.1.1 4.1.1 Primum non nocerePrimum non nocere First, do no harm: Evidence, error and bias 87 First, do no harm: Evidence, error a

15、nd bias 87 4.2 Education and skills 884.2 Education and skills 88 4.2.1 Growing the leaders of the future: Changing education and training 894.2.1 Growing the leaders of the future: Changing education and training 89 4.2.2 Supporting the leaders of today: The need for continuous learning 904.2.2 Sup

16、porting the leaders of today: The need for continuous learning 90 4.3 Data quality, governance, security and interoperability 914.3 Data quality, governance, security and interoperability 91 4.3.1 Digitising health and collecting the right data 914.3.1 Digitising health and collecting the right data

17、 91 4.3.2 Ensuring strong data governance within healthcare organisations 924.3.2 Ensuring strong data governance within healthcare organisations 92 4.3.3 Data interoperability and building bigger datasets 934.3.3 Data interoperability and building bigger datasets 93 4.4 Managing change 944.4 Managi

18、ng change 94 4.5 Investing in new talent and creating new roles 954.5 Investing in new talent and creating new roles 95 4.6 Regulation, policy making and liability, and managing risk 964.6 Regulation, policy making and liability, and managing risk 96 4.6.1 Approving algorithms: Ensuring safety, effi

19、cacy and no bias 974.6.1 Approving algorithms: Ensuring safety, efficacy and no bias 97 4.6.2 Liability: Who is the actor? 1004.6.2 Liability: Who is the actor? 100 4.7 Funding: The most important enabler of all? 1014.7 Funding: The most important enabler of all? 101 4.8 Who should do what? The key

20、roles to make AI happen 1024.8 Who should do what? The key roles to make AI happen 102 4.9 The potential role of the EU 1044.9 The potential role of the EU 104 Chapter 5 Key findings and recommendations 108Chapter 5 Key findings and recommendations 108 5.1 AIs potential to transform healthcare 1095.

21、1 AIs potential to transform healthcare 109 5.2 Status of AI in healthcare internationally 1105.2 Status of AI in healthcare internationally 110 5.3 Impact on healthcare practitioners 1105.3 Impact on healthcare practitioners 110 5.4 Barriers and enablers 1125.4 Barriers and enablers 112 5.5 The imp

22、lications for healthcare organisations and health systems 1145.5 The implications for healthcare organisations and health systems 114 5.6 What role could Europe play? 1155.6 What role could Europe play? 115 Appendix 1: List of interviewees 117Appendix 1: List of interviewees 117 Appendix 2: MGI meth

23、odology 121Appendix 2: MGI methodology 121 Appendix 3: Survey 123Appendix 3: Survey 123 2Transforming healthcare with AI: The impact on the workforce and organisations 3 Artificial intelligence (AI) has the potential to transform how care is delivered. Artificial intelligence (AI) has the potential

24、to transform how care is delivered. It can support improvements in care outcomes, patient experience and It can support improvements in care outcomes, patient experience and access to healthcare services. It can increase productivity and the efficiency access to healthcare services. It can increase

25、productivity and the efficiency of care delivery and allow healthcare systems to provide more and better of care delivery and allow healthcare systems to provide more and better care to more people. AI can help improve the experience of healthcare care to more people. AI can help improve the experie

26、nce of healthcare practitioners, enabling them to spend more time in direct patient care and practitioners, enabling them to spend more time in direct patient care and reducing burnout. Finally, it can support the faster delivery of care, mainly by reducing burnout. Finally, it can support the faste

27、r delivery of care, mainly by accelerating diagnosis time, and help healthcare systems manage population accelerating diagnosis time, and help healthcare systems manage population health more proactively, allocating resources to where they can have the health more proactively, allocating resources t

28、o where they can have the largest impact. largest impact. The implications of introducing and scaling AI in healthcare have been much debated in recent years. The full potential of AI is still being discussed, but questions have been raised about its potential impact on practitioners and certain spe

29、cialties, while issues around ethics, use of personal data and AI-related risks are also being debated. At the same time, healthcare investments in AI are increasing, creating or accentuating disparities in the adoption of innovation in healthcare, and raising questions around the role that health s

30、ystems, public and private players and individual healthcare practitioners can, or should, play in ensuring citizens fully reap the benefits of AI. This joint report between EIT Health and McKinsey and McKinseys Healthcare Systems and Services, and Pharmaceutical and Medical Products practices. The

31、analyses also drew from research on the impact of automation on jobs by MGI, McKinseys independent think tank, established in 1990 to develop a deeper understanding of the evolving global economy and cited as the worlds leading private-sector think tank in the 2018 Global Go To Think Tank Index Repo

32、rt. The research was led by Jorge Fernndez Garca, Director of Innovation, EIT Health; Dr. Angela Spatharou, Partner, Healthcare Systems and Services, McKinsey Jonathan Jenkins, Senior Principal, QuantumBlack; and Solveigh Hieronimus, Partner, McKinsey and from McKinsey Dr. Nicolaus Henke, Senior Par

33、tner and Chairman of QuantumBlack; Dr. Chris Llewellyn, Senior Partner and global leader of Digital and AI services, Pharmaceutical and Medical Products practice; and Dr. Jaana Remes, Partner, MGI. The team was comprised of Maria Fernndez Albizuri, EIT Health and Dr. Mareike Herzog, Dr. Franziska Ga

34、li and Will Taylor, McKinsey Gurneet Singh Dandona, Alok Singh, Shagun Narula and EB Armstrong, MGI; and Paul Mears, Senior Healthcare Advisor, McKinsey in particular, Paul Timmers, Chief Advisor, EIT Health, and Research Associate, University of Oxford, for his extensive feedback on content. Very s

35、pecial thanks are due to Jonathan Turton, our editor, Rich Nunn, Senior Media Designer, Adam Richardson-Foster, Senior Media Designer and Marie Neuhoff, for her support throughout this process. Finally, we would like to extend our deep gratitude to all our interviewees and survey respondents, withou

36、t whose commitment to share their insights and help shape this important topic, this report would not have been possible. Thank you all. Acknowledgements 6 6 Abbreviations A there are ethical debates around how AI and the data that underpins it should be used. This EIT Health and McKinsey a series o

37、f one-to-one interviews with 62 healthcare and other leaders with experience in AI and digital health, and an online survey of 175 healthcare professionals, healthcare investors and AI startup founders and other executives. AI in healthcare being a fast-moving field, the report provides a unique van

38、tage point from the frontline of healthcare delivery and innovation today and the latest view from a wide array of stakeholders on AIs potential, the real state of play today, and what is holding us back. 9Transforming healthcare with AI: The impact on the workforce and organisations Last, to highli

39、ght where AI is already having an impact in healthcare, the report also looks at detailed examples of existing AI solutions in six core areas where AI has a direct impact on the patient and three areas of the healthcare value chain that could benefit from further scaling of AI (Exhibit 1). Exhibit 1

40、 Areas of impact for AI in healthcare In doing so, the report provides a unique contribution to the debate on the impact of AI in healthcare in four ways: 1) decision makers view of the state-of-play in this fast-moving field, where developments from just 12 months ago are considered “old news”; 2)

41、a robust new methodology to evaluate the impact of automation and AI on specific skills and activities in healthcare in Europe; 3) a substantial review of use cases that illustrate the potential that AI is already on track to deliver; 4) a unique view from the frontline, hearing from healthcare prof

42、essionals, investors and startup executives on where the real potential, opportunities and barriers lie. The report does not attempt to cover all facets of this complex issue, in particular the ethics of AI or managing AI-related risks, but does reflect the efforts on this important topic led by EIT

43、 Health and other EU institutions. Equally, while it acknowledges the potential disruptive impact of personalisation on both healthcare delivery and healthcare innovation in the future (e.g., in R and others even help optimise healthcare R while the healthcare professionals saw the private sectors r

44、ole in areas such as aggregating or analysing data, providing a secure space for data lakes, or helping upskill healthcare staff as minimal or nonexistent. One problem AI solutions face is building the clinical evidence of quality and effectiveness. While startups are interested in scaling solutions

45、 fast, healthcare practitioners must have proof that any new idea will “do no harm” before it comes anywhere near a patient. Practitioners also want to understand how it works, where the underlying data come from and what biases might be embedded in the algorithms, so are interested in going past th

46、e concept of AI as a “black box” to understand what underpins it. Transparency and collaboration between innovators and practitioners will be key in scaling AI in European healthcare. User-centric design is another essential component of a quality product. Design should have the end user at its hear

47、t. This means AI should fit seamlessly with the workflow of decision makers and by being used, it will be improved. Many interviewees agreed that if AI design delivers value to end users, those users are more likely to pay attention to the quality of data they contribute, thereby improving the AI an

48、d creating a virtuous circle. Finally, AI research needs to heavily emphasise explainable, causal and ethical AI, which could be a key driver of adoption. 2) Rethinking education and skills. We have already touched on the importance of digital skills these are not part of most practitioners arsenal

49、today. AI in healthcare will require leaders well-versed in both biomedical and data science. There have been recent moves to train students in the science where medicine, biology and informatics meet through joint degrees, though this is less prevalent in Europe. More broadly, skills such as basic digital literacy, the fundamentals of genomics, AI and machine learning need to become mainstream for all practitioners, supplemented by critical-thinking skills and the development of a continuous- learning mindset. Alongside upgrading clinical training