AI use in European - HIMSS
Transcript of AI use in European - HIMSS
1
© HIMSS Analytics | eHealth TRENDBAROMETER May 2018
1www.himss.eu/analytics
„Artificial Intelligence (AI)“
Results, May 2018
AI use in European
healthcare
2
© HIMSS Analytics | eHealth TRENDBAROMETER May 2018
TABLE OF CONTENT
• Introduction & methodology page 3
• Survey questions page 4
• Results
Key Findings page 5
Used and planned Artificial Intelligence (AI) tools page 7
Budget for AI investments page 11
Challenges for AI in healthcare page 12
Benefits from AI in healthcare page 13
Breakthrough for AI tools in healthcare page 17
Most advanced country with regards to the use of AI in healthcare page 18
Business expectations page 19
3
© HIMSS Analytics | eHealth TRENDBAROMETER May 2018
HIMSS ANALYTICS – WHO WE ARE
HIMSS Analytics in Europe provides healthcare organisations, governments and industry with extensive data resources and services about the adoption and use of healthcare IT in Europe. HIMSS Analytics’ offerings include database and advisory solutions which encompass market research, IT adoption benchmarking, IT Maturity Models for topics like Electronic Medical Records or Continuity of Care. These offerings are designed to support Management and CIOs, IT Executives and Clinicians from across Europe to compare and measure their progress.
EHEALTH TRENDBAROMETER – METHODOLOGY
Objectives o Continuous evaluation of trends and issues in the European eHealth sector:2 – 4 survey waves per year, with both varying and recurring topics
o Provide insights into current and desired states of eHealth in Europe
o Enable discussions within the European eHealth community
Study design o Structured quantitative online survey
o Quick completion (time-to-complete is < 5 min)
o Survey language(s): English, German
o Participation via personal email invitation or via public link on www.himss.eu and other channels
Target audienceand participants
o Key audience: eHealth professionals from several European countries, especially:
o IT staff, administrative staff, and clinicians from health facilities (e.g. CIO’s, CEO’s, Physicians, Nurses)
o Professionals from health-IT related software and consulting companies
o Professionals from other eHealth related sectors (e.g. health authorities, research, journalism)
o Number of participants: approx. 300 – 500 per survey wave
Survey period o Typical field time: Approx. 2 months
o Specific field time for „Artificial Intelligence“ TRENDBAROMETER:
February – March 2017
Access all eHealth TRENDBAROMETERS here:
www.himss.eu/healthcare-providers/ehealth-trends
4
© HIMSS Analytics | eHealth TRENDBAROMETER May 2018
SURVEY QUESTIONS
1. What type of organisation are you working for?
2. Does your organisation already use or do you have specific plans to implement/provide Artificial Intelligence (AI) tools?
3. When will the planned AI tools be implemented/launched?
4. In which areas does your organisation use/have or plan to implement/provide AI tools?
5. Do you think your organisation provides sufficient budget to invest in AI over the next 12 months?
6. In your opinion, which are the biggest challenges for AI in healthcare?
7. In your opinion, which are the biggest benefits from using AI in healthcare?
8. When do you expect a breakthrough for AI tools in healthcare, i.e. when will they become powerful, useful and widespread in your country?
9. In your opinion, which country has made most progress so far with regards to the use of AI in healthcare?
10. How will the environment for eHealth innovation and investment in your country develop over the next 12 months?
Main Topic: „Artificial Intelligence“
Artificial Intelligence in this study is defined as…… computer applications designed to sense, understand, act and learn so that they can perform
administrative and clinical healthcare functions, either on their own or to augment human activities. Typically, this would include the use of deep learning technology.
5
© HIMSS Analytics | eHealth TRENDBAROMETER May 2018
KEY FINDINGS
Lesson 1: Healthcare providers are just starting to use AI tools.
Only 16% of healthcare facilities in Europe currently use AI tools (of which they are aware). In general the use of AI tools in health facilities appears to trail significantly behind what the industry is already offering. The variety of available AI tools is pretty broad, with solutions that provide better workflow support slightly dominating. The Nordic countries and the Netherlands are European pioneers in the use of AI tools.
Lesson 2: AI for healthcare is a growing market expected to breakthrough in 5 years.
Get ready to see AI tools being used all over the place in about 5 years. eHealth professionals expect them to be powerful, useful and widespread by 2023. In the meantime, 21% of health facilities have specific purchase plans for AI tools, every second software vendor plans new product launches in that field, and nearly all of this will be executed within the next 3 years. However, health facilities remain cash-strapped, with a majority not having sufficient budgets for AI, at least in the near term.
Lesson 3: AI tools must mature and confidence in products needs to be increased.Employees from health facilities and governmental authorities are concerned about the maturity of currently existing AI tools. Software vendors are aware of those concerns and will to face this lack of trust from clinicians and other potential customers. AI tools also need to prove that they can handle data privacy and are either providing a clear return on investment or can be implemented and used relatively inexpensively.
Lesson 4: This is no hype! eHealth professionals, users and producers, see a world of opportunities for AI use in healthcare.eHealth professionals appear very confident that AI will help to improve care quality, support clinicians to take better and more informed decisions, provide more personalised treatments, enable healthcare professionals to save time, be more efficient, focus on complex or urgent cases etc. Software vendors in particular also highlight the potential that AI will help to reduce medical errors, raise efficiency and enable cost savings – which needs to be proven.
Lesson 5: The US is considered the most advanced country in terms of using AI in healthcare.The European eHealth community has the impression that the US is the key market for AI tools and usage at the moment. A mature IT infrastructure in health-care, a large market with 325 million citizens and the fact that the most successful and innovative software vendors and venture capitalists are located there, might stimulate that impression. In Europe, the Nordic countries (incl. Estonia) are perceived to be most advanced to use AI tools.
eHealth Business climate: The business trend for eHealth continues to be positive, albeit at a slightly lower level than at the end of 2017.
Lessons on Artificial Intelligence (AI) in healthcare5
6
© HIMSS Analytics | eHealth TRENDBAROMETER May 2018
Software vendor• CEO (18.9%)• Organisational and Corporate
Governance (15.6%)• IT Staff (7.8%)• Researcher/Scientist (7.8%)• Software Developer (5.6%)• CIO (4.4%)• Quality Management Staff (3.3%)
• Other position (86.7%)
Other• Researcher/Scientist (32.6%)• CEO (14.9%)• Organisational and Corporate
Governance (8.5%)• IT Staff (4.3%)• Quality Management Staff (3.5%)• CIO (2.1%)• Other position (34.0%)
20%
7%
12%
8%16%
7%
9%
21%
Germany
Austria
Switzerland
Netherlands
Nordic Countries
Italy
Spain
Other
Governmental health authority• Organisational and
Corporate Governance (26.6%)
• CIO (17.2%)• IT Staff (15.6%)• Researcher/Scientist (10.9%)• Quality Management Staff (6.3%)• CEO (3.1%)• Clerk (1.6%)• Other position (18.8%)
SAMPLE DISTRIBUTION – SURVEY PARTICIPANTS
Health facility• CIO (21.5%)• Medical Profession (Physician) (21.5%)• IT Staff (15.2%)• Medical Profession (Nurse, Pharmacist...)
(5.9%)• Organisational and Corporate
Governance (5.6%)• CMO (4.8%)• CEO (4.8%)• Quality Management Staff (4.4%)• Researcher/Scientist (4.4%)• Other position (11.9%)
Geographic distributionRespondent‘s occupation
n=562
Countries or regions with more than 30 participants are shown individually.
*„Other“ respondents are from: Greece,Turkey, Estonia
Country n
Germany 121
Switzerland 65
Spain 50
Netherlands 44
Austria 42
Other* 41
Italy 37
Denmark 32
Belgium 25
Sweden 22
Norway 20
United Kingdom 17
France 16
Finland 15
Ireland 15
Total 562
n=562
7
© HIMSS Analytics | eHealth TRENDBAROMETER May 2018
USE OF AI TOOLS – BY TYPE OF ORGANISATION
Does your organisation already use or do you have specific plans to implement/provide Artificial Intelligence (AI) tools?[Only participants who are working in a healthcare facility or for a SW vendor]
The vast majority of health professionals in Europe (84%) currently does not use AI (Artificial Intelligence) tools, or are at least not aware of it. One out of four health professionals is aware about specific plans to implement AI tools over the next 1 – 3 years. Software vendors are a bit ahead of the curve. More than one out of four (28%) already offer AI tools, and another 51% have specific plans to launch such products. We expect this to become a very competitive market over the next few years due to many new solution providers entering that sector, many of them being start-ups or established companies from non-healthcare related industry segments.
8
© HIMSS Analytics | eHealth TRENDBAROMETER May 2018
USE OF AI TOOLS – BY COUNTRY OR REGION
Does your organisation already use or do you have specific plans to implement Artificial Intelligence (AI) tools?[Only participants who are working in a healthcare facility; total including „other“]
Healthcare providers in the Nordic countries and the Netherlands are at the forefront of AI usage in Europe. At the bottom of the table are the D-A-CH countries, with just 5% to 10% of health professionals currently using AI tools. When looking at plans to purchase and implement AI tools it also appears that healthcare professionals in Germany and Austria are more cautious than their peers in other countries to use such products in the near future. However, in general the demand for AI tools is certainly increasing, especially in Switzerland and Spain where between 37% to 42% of health professionals claim to have purchase plans for AI solutions.
Nordic Countriesn=29
Netherlandsn=18
Italyn=16
Spainn=19
Switzerlandn=30
Germanyn=68
Austrian=21
Totaln=249
9
© HIMSS Analytics | eHealth TRENDBAROMETER May 2018
PLANS AND BUDGETS FOR AI TOOLS – BY TYPE OF ORGANISATION
When will the planned AI tools be implemented/launched?[Only participants who are working in a healthcare facility or for a SW vendor and who have plans to implement/launch AI tools]
If plans exist to implement or launch AI tools, most of them will be realised within the next 2 years. This is similar for healthcare providers as for software vendors. In general, more than 90% of already existing plans will be realised within a 3-year period, i.e. by the year 2021. Despite the interest and will to purchase AI tools, lack of funds can be a roadblock for healthcare providers, at least in the short term. 34% of those that already have plans claim that their organisation doesn’t provide a sufficient budget to invest in AI over the next 12 months.
… of health facility respondents have specific plans to
purchase AI tools
34%
66%
Yes No
32%
37%
23%
9%
37%
39%
17%
7%
Within 1 year
1 - 2 years
2 - 3 years
More than 3 years
Health facility (n=57) Software vendor (n=41)
25%
… of software vendor respondents have specific plans to launch AI tools
51%
Do you think your organisation provides sufficient budget to invest in AI over the next 12 months?[Only participants who are working in a healthcare facility and who already use or plan to implement AI tools; total including „other“]
n=83Healthcare providers
only
10
© HIMSS Analytics | eHealth TRENDBAROMETER May 2018
TYPE OF AI TOOLS – BY TYPE OF ORGANISATION
In which areas does your organisation use/have or plan to implement/provide AI tools?[Only participants who are working in a healthcare facility or IT software vendor and who already use/have or plan to implement/provide AI tools]
At the moment, healthcare providers use AI tools most frequently for Workflow Assistance (14%) and in Research (13%), closely followed by Medication Administration (12%) and Radiology (11%). Those areas, plus Oncology, are also where healthcare providers have most of their AI investment plans. A similar pattern is visible when looking at what software vendors already offer, or plan to launch in the near future. AI tools that provide better workflow support for clinicians or offer a preliminary diagnosis stand a bit out and appear to be exceptionally promising for software vendors.
11
© HIMSS Analytics | eHealth TRENDBAROMETER May 2018
Overall, a lack of product maturity and trust from medical staff are perceived to be the biggest challenges for a more widespread use of AI in healthcare. This perception differs quite significantly by type of organisation. Software vendors on the one hand are currently mostly challenged by having insufficient access to high quality data to train their AI tools as well as missing trust from medical staff in their tools. Health facility employees on the other hand are cautious to use or purchase AI tools because they think they are immature, too expensive and there’s a lack of knowledge about their availability and capabilities. Governmental health authorities tend to be more concerned about gaps regarding to the interoperability of those tools (15%).
CHALLENGES FOR AI IN HEALTHCARE – BY TYPE OF ORGANISATION
In your opinion, which are the biggest challenges for AI in healthcare?
Insufficient maturity of
the products available
on the market
Lack of trust from
medical staff
Data Privacy Interope-rability with other soft-/hardware
Lack of legal approval to
use AI applications
Lack of sufficient
high quality data to train
AI tools
Insufficient user
education/ knowledge
High costs Lack of trust from
patients
Implemen-tation of
new software
[multiple responses possible]
12
© HIMSS Analytics | eHealth TRENDBAROMETER May 2018
CHALLENGES FOR AI IN HEALTHCARE – BY COUNTRY OR REGION
In your opinion, which are the biggest challenges for AI in healthcare?[Total including „other“; overall top 2 answers; multiple responses possible]
Italian eHealth professionals identify legal approval issues (21%) and lack of trust (20%) from medical staff as biggest roadblocks for more widespread use of AI solutions. Professionals from the Nordics are waiting for more mature AI solutions (18%). Dutch professionals are challenged by data privacy regulations (17%). Germans are concerned about high costs (13%).
Ge
rman
y
Au
stri
a
Swit
zerl
and
Ne
the
rlan
ds
No
rdic
C
ou
ntr
ies
Ital
y
Spai
n
Tota
l
Insufficient maturity of the products available on the market
13% 14% 11% 8% 18% 9% 11% 13%
Lack of trust from medical staff 10% 16% 11% 15% 10% 20% 16% 13%
Data Privacy 10% 15% 13% 17% 16% 7% 12% 12%
Interoperability with other soft-/hardware
12% 10% 13% 15% 10% 5% 13% 11%
Lack of legal approval to use AI applications
11% 11% 11% 8% 12% 21% 12% 11%
Lack of sufficient high quality data to train AI tools
9% 8% 11% 8% 13% 13% 15% 11%
Insufficient user education/knowledge
11% 12% 13% 6% 8% 11% 8% 11%
High costs 13% 5% 7% 7% 5% 8% 6% 8%
Lack of trust from patients 5% 5% 6% 5% 2% 2% 4% 4%
Implementation of new software 2% 3% 2% 5% 3% 1% 4% 3%
N answers 334 118 174 111 220 85 137 1490
High Costs (14%)
Insufficient maturity of the products available on the market (13%)
Lack of trust from medical staff (16%)
Data Privacy (15%)
Lack of legal approval to use AI applications (21%)
Lack of trust from medical staff (20%)
Lack of trust from medical staff (16%)
Lack of sufficient high quality data to train AI tools (15%)
Data Privacy (13%)
Insufficient usereducation/knowledge / Interoperability with other soft-/hardware (13%)
Data Privacy (17%)
Lack of trust from medical staff / Interoperability with other soft-/hardware (15%)
Insufficient maturity of the products available on the market (18%)
Data Privacy (16%)
13
© HIMSS Analytics | eHealth TRENDBAROMETER May 2018
BENEFITS FROM USING AI IN HEALTHCARE – OVERALL RESULTS
In your opinion, which are the biggest benefits from using AI in healthcare?
The advent of new AI solutions, many of them based on artificial neural networks and deep learning, trigger a lot of optimism to improve health outcomes and enable new ways to deliver healthcare services. Many statements from eHealth professionals circle around the hope that AI will help to improve care quality, support clinicians to take better and more informed decisions, provide more personalised treatments, enable healthcare professionals to save time, be more efficient, focus on complex or urgent cases etc. While there is a lot of enthusiasm, some eHealth professionals also raise their concerns about AI currently being a hype and that real benefits are, at best, to be seen in many years.
[free text responses]
14
© HIMSS Analytics | eHealth TRENDBAROMETER May 2018
14
It will enable each citizen drive their health and well-being, to improve the overall quality, access and cost effectiveness of our healthcare services.
AI offers the possibility to consider the complete knowledge about healthcare and learn from
previous experiences.
AI should help to identify “the abnormal" to
healthcare professionals, and reduce their
workload by helping them to discard
everything that is normal.
AI will help to overcome boundaries between health care facilities, early detection or prognosis of diseases,
support doctors in diagnosis and therapy.
It will help with the optimization and standardization of procedures, support and improvement of diagnostic and therapeutic
safety
AI will offer support for clinical decision
making by analyzing large
volumes of information, from
EMR and any other sources of
information.
I think AI is over hyped in Health, it will take long
time to achieve cases that will give benefit for
patients.
We will gain new insights and knowledge, deliver more care with improved quality, put the patients in the drivers seat, focus human
resources on complex tasks.
Apart from settings that require creative thinking and rational, AI
obviously trumps human memory and diversity of knowledge.
AI will allow us to do tasks much more efficiently and around the clock - like reading Radiology -done by costly professionals who work on a 9-5
basis.
BENEFITS FROM USING AI IN HEALTHCARE – SELECTED QUOTES
Selected quotes related to the question: “In your opinion, which are the biggest benefits from using AI in healthcare?”
15
© HIMSS Analytics | eHealth TRENDBAROMETER May 2018
BENEFITS FROM USING AI IN HEALTHCARE – BY TYPE OF ORGANISATION
In your opinion, which are the biggest benefits from using AI in healthcare?
AI is expected to improve the quality of care. Especially professionals working in health facilities (21%) see this as a key benefit. Those professionals think that, if rightly used, AI will support better medical decisions and improve diagnostics. Software vendor professionals on the other hand focus more on error reduction, efficiency gains and cost savings when they think about AI benefits. Health authority representatives perceive AI to be very beneficial by helping to process large amounts of data (to support clinicians to take better medical decisions).
[free text responses, categorised by HIMSS]
16
© HIMSS Analytics | eHealth TRENDBAROMETER May 2018
BENEFITS FROM USING AI IN HEALTHCARE – BY COUNTRY OR REGION
In your opinion, which are the biggest benefits for AI in healthcare?[Total including „other“; overall top 2 answers; free text responses, categorised by HIMSS]
In general country differences are rather nuanced. Italian eHealth professionals appear to have the biggest expectations regarding AI helping to improve care outcomes and supporting better clinical decision making. In Austria there’s little hope that AI will contribute to decrease healthcarecosts or leads to time savings. Swiss and Nordic professionals perceive to gain value from AI by processing large amounts of (patient) data.
Improve quality ofcare (22%)
Improve diagnostics (13%)
Improve quality of care (23%)
Improve medical decisionmaking / Improve diagnostics/ Standardize processes (10%)
Improve medical decision making (25%)
Improve quality of care (21%)
Improve medical decision making (25%)
Improve quality of care (13%)
Improve quality of care (18%)
Process large amount of data(15%)
Improve quality of care (19%)
Save time (13%)
Improve quality of care (21%)
Process large amount of data (16%)
Ge
rman
y
Au
stri
a
Swit
zerl
and
Ne
the
rlan
ds
No
rdic
C
ou
ntr
ies
Ital
y
Spai
n
Tota
l
Improve quality of care 22% 23% 18% 19% 21% 21% 13% 19%
Improve medical decission making 9% 10% 10% 11% 12% 25% 25% 13%
Improve diagnostics 13% 10% 8% 7% 8% 8% 10% 10%
Process large amount of data 8% 8% 15% 7% 16% 4% 10% 10%
Lower costs 9% 3% 10% 9% 3% 9% 8% 8%
Save time 11% 5% 10% 13% 8% 4% 6% 8%
Reduce medical errors 4% 8% 7% 7% 2% 8% 10% 7%
Standardize processes 4% 10% 7% 4% 3% 6% 2% 5%
Generate new knowledge 3% 0% 3% 9% 8% 4% 6% 5%
Improve precision medicine 2% 3% 0% 6% 3% 4% 4% 3%
Relief of personnel 7% 3% 5% 0% 3% 2% 0% 3%
N answers 116 39 60 54 97 53 52 619
17
© HIMSS Analytics | eHealth TRENDBAROMETER May 2018
YEAR OF BREAKTHROUGH FOR AI TOOLS IN HEALTHCARE
When do you expect a breakthrough for AI tools in the healthcare space, i.e. when will they become powerful, useful and widespread in your country?
Get ready for widespread use of AI in 5 years! According to their own estimation, eHealth professionals in Europe will routinely and effectively use Artificial Intelligence tools by 2023. Respondents from southern Europe and the Netherlands are a bit more enthusiastic and expect a breakthrough by 2022 while their peers from Austria or Switzerland are a more cautious see it happen 2 years later.
Country n Average estimated year of AI
breakthrough
Netherlands SpainItaly
414535
2022
GermanyNordic CountriesOther Countries
10382
1072023
AustriaSwitzerland
3555
2024
Total 503 2023
Op
tim
isti
c co
un
trie
sC
auti
ou
s co
un
trie
s
Average year ofAI breakthrough
2023
Grey line: percent of respondents indicating a specific year of AI breakthroughColumns: cumulated percentage of respondents (i.e. no respondent indicated a year after 2028)
18
© HIMSS Analytics | eHealth TRENDBAROMETER May 2018
MOST ADVANCED COUNTRY WITH REGARDS TO THE USE OF AI – BY TYPE OF ORGANISATION
In your opinion, which country has made most progress so far with regards to the use of AI in healthcare?[Total including „other“]
The vast majority sees the US as a pioneer in the field of using AI in healthcare (Total: 52%). Estonia was rated as the most advanced European country (8%), especially from the perspective of health facility employees. Software vendors rather see Denmark and Finland in the lead in Europe when it comes to the use of AI.
Totaln=308
Health facilityn=136
IT Software vendorn=50
Governmental health authorityn=36
19
© HIMSS Analytics | eHealth TRENDBAROMETER May 2018
28 24
42 4338 35
49
35
7469
77 73 73
60
89
65 6960
7075
65
27
50
36
-6
28 30
57 52 54 48 5346
1 2 3 4 5 6 7 8
DACH Nordic Countries Netherlands UK All countries
BUSINESS EXPECTATIONS
From a general perspective: How will the environment for eHealth innovation and investment in your country develop over the next 12 months?[Score for “Balance of business expectations” = (percentage “improve” – percentage “worse”)*100]
+100-100
Bal
ance
of
bu
sin
ess
exp
ecta
tio
ns*
Very positive expectations
Very negative expectations
2015 – Q2 2015 – Q3 2016 – Q2 2016 – Q3 2016 – Q4
Business prospects for the eHealth sector continue to be positive, albeit at a slightly lower level than at the end of 2017. Dutch eHealth professionals are particularly optimistic, 77% expect an improvement and only 2% believe the situation will become worse. The dip seen last year in the UK appears to have been temporarily only, with ratings having picked up this time. However, due to the relative low response from UK participants in the current survey wave this needs to be interpreted carefully.
worse steady improve N
Germany 10% 40% 50% 121
Austria 12% 50% 38% 42
Switzerland 11% 42% 48% 65
Netherlands 2% 20% 77% 44
Belgium 0% 32% 68% 25
Denmark 13% 38% 50% 32
Norway 5% 25% 70% 20
Sweden 0% 27% 73% 22
Finland 0% 20% 80% 15
United Kingdom 18% 24% 59% 17
Italy 11% 51% 38% 37
Spain 10% 43% 47% 49
All countries 9% 37% 54% 562
Results by country (2018 – Q1)
2017 – Q2 2017 – Q3 2018 – Q1
20
© HIMSS Analytics | eHealth TRENDBAROMETER May 2018
Thank you for your participation!
HIMSS Analytics
Office Leipzig
Schwägrichenstraße 9
04107 Leipzig
Germany
www.himss.eu/analytics
www.himssanalytics.org
Join the European eHealth community panel