Online symptom checkers for self-diagnosis and triage

by Maria Voukelatou (2046993)

Health, Care, Technology & Regulation

Professors Tineke Broer
Tilburg University
Law & Technology
Tilburg, May 29, 2020
 Online symptom checkers for self-diagnosis and triage

Table of Contents
1. Introduction................................................................................................................3
1.1. Focus.......................................................................................................................3
1.2 Regulatory Challenge...............................................................................................3
2. Regulation of Online Symptom Checkers..................................................................4
2.1. Applicable Laws......................................................................................................4
2.2. Legal Gaps in Regulation........................................................................................6
3. Ethical Issues..............................................................................................................7
4. Socio-political Consequences of using Online Symptom Checkers..........................8
5. Socio-political Consequences of Different Regulatory Choices................................9
5.1. No Regulation.........................................................................................................9
5.2. Prohibit online symptom checkers..........................................................................9
5.3. Regulating online symptom checkers without safeguards....................................10
6. Providers’ Contribution to governance and use of online symptom checkers.........10
7. Conclusion................................................................................................................11

2
 Online symptom checkers for self-diagnosis and triage

1. Introduction

1.1. Focus
This paper focuses on online symptom checkers for self-diagnosis and triage and their
legal, ethical, and socio-political challenges within the European Union (EU). An
online symptom checker is an Artificial Intelligence (AI) powered software,
accessible via a website or mobile application. Online symptom checkers request
several information about users’ symptoms using computerized algorithms or allow
users to enter information about their symptoms themselves 1. Symptom checkers have
intertwined function: to enable self-diagnosis and to help with triage 2. The self-
diagnosis feature provides a listing of diagnoses, usually ranked by likelihood, while
the triage function tells patients whether, where and with what urgency they should be
seeking treatment3.
Online symptom checkers are capable of diagnosing several diseases, from a simple
flu to cancer and mental issues, like depression4. Recently, the majority of the existing
online symptom checkers added an extra module into their diagnostic platforms
concerning COVID-19. This feature evaluates whether the user should visit a doctor,
get isolated or checked, like in any other case.

1.2 Regulatory Challenge

As technology raises new possibilities and people engage in new forms of behavior,
the law continues to aim at fixing old problems and is sometimes not able to 'keep up'
with the contemporary world5. The difficulty of adapting legal frameworks to changes
is frequently depicted as “law” losing a race, the so-called pacing problem6.

One aspect of newness that is important from a regulatory standpoint when it comes
to emerging technologies, such as online symptom checkers, is the so-called
Collingridge dilemma7. D. Collingridge argued that regulators responding to a new
technology face the following hurdles8: “At an early stage in a technology’s
development, regulation was problematic due to the lack of information about the
technology’s likely impact. At a later stage, regulation was problematic as the
1
Semigran, H. L. (2015). Evaluation of symptom checkers for self-diagnosis and triage: audit study.
Bmj. British Medical Journal, 8016, 11–11.
2
Ibid.
3
Ibid.
4
A to Z Guides, WebMD, https://www.webmd.com/a-to-z-guides/health-topics (Last Access: May 15,
2020)
5
Bennett Moses, Lyria, Agents of Change: How the Law ʻCopesʼ with Technological Change (January
27, 2012). Griffith Law Review, Vol. 20, No. 4, pp. 764-794, 2011; UNSW Law Research Paper No.
2012-2. Available at SSRN: https://ssrn.com/abstract=2000428 (Last Access: April 2, 2020)
6
Ibid
7
Bennett Moses, Lyria, How to Think About Law, Regulation and Technology: Problems with
'Technology' as a Regulatory Target (2013). (2013) 5(1) Law, Innovation and Technology 1-20;
UNSW Law Research Paper No. 2014-30. Available at SSRN: https://ssrn.com/abstract=2464750 (Last
Access: April 2, 2020)
8
Ibid

3
 Online symptom checkers for self-diagnosis and triage

technology would become more entrenched, making any changes demanded by

regulators expensive to implement9.”
Thus, regulators deal with an “uncertainty paradox” since without sufficient risk
information of technological developments, they are forced to make decisions10. There
may be partial solutions to this problem, like adopting a specific approach, such as the
precautionary principle (PP)11. The PP aims to predict and mitigate potentially severe
or irreversible risks under scientific uncertainty, thereby maintaining the potential for
future technological advances and their spread12.
Another aspect of newness that is important from a regulatory point of view when it
comes to emerging technologies, such as online symptom checkers, is the so-called
Innovation Principle (IP). IP is a medium for helping the EU to achieve its policy
goals by ensuring that legislation is structured to establish the best possible conditions
for innovation to thrive 13.
The regulatory problem that we have to deal with since online symptom checkers
already exist, is not whether to regulate or not, but how exactly to regulate a safe
environment for symptom checking and triage, in order precautionary and innovation
principles would be used alongside each other, recognizing the need to protect human
rights while also protecting Europe’s ability to innovate14.

2. Regulation of Online Symptom Checkers

2.1. Applicable Laws

Currently, fragmentary legislation is relevant to online symptom checkers for self-
diagnosis and triage rather than a sui generis regulatory regime.
The online symptom checkers can be classified under Class I according to Annex VIII
Chapter 3 (4.1) of Regulation (EU) 2017/745 on medical devices (MDR) 15 as a
noninvasive device. Under MDR, manufacturers have obligations to monitor the

9
Ibid
10
Ibid
11
Ibid
12
Som, C., Hilty, L.M. & Köhler, A.R. The Precautionary Principle as a Framework for a Sustainable
Information Society. J Bus Ethics 85, 493 (2009). https://doi-org.uow.idm.oclc.org/10.1007/s10551-
009-0214-x (Last Access: April 2, 2020)
13
Ensuring EU legislation supports innovation, https://ec.europa.eu/info/research-and-innovation/law-
and-regulations/innovation-friendly-legislation_en (Last Access: April 2, 2020)
14
Kathleen Garnett, Geert Van Calster & Leonie Reins (2018) Towards an innovation principle: an
industry trump or shortening the odds on environmental protection?, Law, Innovation and Technology,
10:1, 1-14, DOI: 10.1080/17579961.2018.1455023 (Lase Access: April 2, 2020)
Regulation (Eu) 2017/745 of the European Parliament And of the Council of 5 April 2017 on
15

medical devices, amending Directive 2001/83/EC, Regulation (EC) No 178/2002 and Regulation (EC)
No 1223/2009 and repealing Council Directives 90/385/EEC and 93/42/EEC.

4
 Online symptom checkers for self-diagnosis and triage

quality, safety, and performance of devices and in case of a defective device they can
be held liable under the Product Liability Directive16.
However, as previously stated, a symptom checkers is an AI powered software that
uses machine learning17 and natural language processing18 technology to perform its
final output. That means that the software-device despite the fact that it might not be
defective, can potentially generate an undesirable decision due to its self-learning
ability. Unfortunately, so far, the EU has no legislation on AI. Instead, the European
Commission (EC) has published several non-binding documents such as -inter alia-
the White Paper, the Report on the Safety and Liability aspects of AI, or the Ethics
Guidelines for Trustworthy AI. These documents are considered as the foundation of
the EC’s coordinate effort to issue binding legislative acts.
Recently the EC issued the “Guidance on Apps supporting the fight against COVID
19 pandemic in relation to data protection”19. Symptom checkers, according to the EC,
fall within the spectrum of the Guidance.
Under Article 5 of the ePrivacy Directive20, storing of -the required for the function of
the symptom checker application- information on the user’s device or gaining access
to the information already stored is allowed only if (i) the storage and/or access is
strictly necessary for the information society service explicitly requested by the user,
or if (ii) the user has given consent. This consent should be “freely given”, “explicit”,
“specific”, and “informed” within the meaning of the General Data Protection
Regulation (GDPR)21.
The EC, also stresses, that the processing of personal data and particularly of special
categories of personal data by National Health authorities, may take place under the
conditions of Article 6(1)(c), 6(3) and Article 9(2)(i) of the GDPR or when such

16
Council Directive 85/374/EEC of 25 July 1985 on the approximation of the laws, regulations and
administrative provisions of the Member States concerning liability for defective products.
17
Machine learning is one of AI’s subdomains and has been defined in different ways and can be
summarized as “the changes in machine programming or data that occurs in order to improve machine
performance”. (Brown, Benjamin. Artificial Intelligence: 2 Books In 1:The Beginner’s Guide to Data
Science, Data Analytics, Deep Learning, Machine Learning, Mining, Deep Medicine. The Revolution
of the Future Today!. 584)
18
Natural language processing, means “getting computers to deal with human language.” Natural-
language processing includes topics such as speech recognition, web search, automated question
answering, and machine translation. (Mitchell, Melanie. Artificial Intelligence. Farrar, Straus, and
Giroux. 178) Natural language processing is the ability of technology to interact with human language,
especially in terms of perceiving and interpreting language in a natural, fluent way. (Brown, Benjamin.
Artificial Intelligence: 2 Books In 1:The Beginner’s Guide to Data Science, Data Analytics, Deep
Learning, Machine Learning, Mining, Deep Medicine. The Revolution of the Future Today! . 1855)
19
European Commission. Guidance on Apps supporting the fight against COVID 19 pandemic in
relation to data protection (2020/C 124 I/01). https://eur-lex.europa.eu/legal-content/EN/TXT/PDF/?
uri=CELEX:52020XC0417(08)&from=EN (Last Access: May 8, 2020)
20
Directive 2002/58/EC of the European Parliament and of the Council of 12 July 2002 concerning the
processing of personal data and the protection of privacy in the electronic communications sector
(Directive on privacy and electronic communications) (OJ L 201, 31.7.2002, p. 37)
21
European Commission. Guidance on Apps supporting the fight against COVID 19 pandemic in
relation to data protection (2020/C 124 I/01). https://eur-lex.europa.eu/legal-content/EN/TXT/PDF/?
uri=CELEX:52020XC0417(08)&from=EN (Last Access: May 8, 2020)

5
 Online symptom checkers for self-diagnosis and triage

processing is required for the performance of a task carried out to further the public
interest recognized by EU or Member State law22.

2.2. Legal Gaps in Regulation

As previously stated, there is no regulatory framework for AI, thus someone could
consider this fact as a constrain to the technological advances in medicine related to
AI applications. One of the main concerns in relation to symptom checkers is the legal
liability issue. The imputation of potential liability arising from the usage of online
symptom checkers may be complicated since several actors are involved from the
developing of the software to its final output. The traditional doctor-patient
relationship is jeopardized, since there is no doctor involved in the diagnosis and
triage procedures. Instead, the doctor’s tasks are substituted by an algorithm that
provides advices according to the data input.
A wrong diagnosis or triage could lead to the damage of patient’s health. On a
national level, it can be observed that Member States' legislation does not yet include
clear liability rules related to damages arising from the use of emerging digital
technologies, such as AI in the health care sector 23. Τhe harmful effects of the
operation of emerging digital technologies can be compensated under existing
traditional laws on damages in tort in each Member State 24. Τhese regimes may not
always lead to satisfactory and adequate results nor lead to the desired degree of
uniformity of outcomes25.
The characteristics of emerging digital technologies like AI, challenge aspects
of Union and national legal frameworks and could reduce their effectiveness 26.
AI in medicine may cause more unprecedented difficulties, as it may be more difficult
to determine what caused the damage in some circumstances, particularly if the AI
algorithm can learn new things by itself 27. These new risks are common for all EU
Member States and must be unitedly addressed.
Various types of AI and its diverse domains of application, such as online symptom
checkers for self-diagnosis and triage, raise particular ethical and regulatory human
rights issues. In order to guarantee that they protect people from the risks caused by
22
Regulation (EU) 2016/679 of the European Parliament and of the Council of 27 April 2016 on the
protection of natural persons with regard to the processing of personal data and on the free movement
of such data, and repealing Directive 95/46/EC (General Data Protection Regulation) (OJ L 119,
4.5.2016, p. 1)
23
Ibid.
24
Ibid.
25
Ibid.
26
Commission to the European Parliament, the Council and the European Economic and Social
Committee (2020). Report on the safety and liability implications of Artificial Intelligence, the Internet
of Things and robotics. https://ec.europa.eu/info/sites/info/files/report-safety-liability-artificial-
intelligence-feb2020_en_1.pdf (Last Access: April 29, 2020)
27
European Parliament’s Legal Affairs Committee (2017). European civil law rules in robotics.
https://www.europarl.europa.eu/RegData/etudes/STUD/2016/571379/IPOL_STU(2016)571379_EN.pd
f (Last Access: April 29, 2020)

6
 Online symptom checkers for self-diagnosis and triage

AI, existing laws must be re-evaluated, and if necessary revised, to tackle the new and
emerging threats. Governments need to discuss how current legislation can be
extended to AI, identify areas, sectors and use cases where relevant regulations need
to be revised 28.

3. Ethical Issues

One could hardly gather all the ethical issues, concerning online symptom checkers
for self-diagnosis and triage in one paper, since not all the risks of this technology are
known. This paper only touches upon some of the most important ethical challenges.

Safety and health is one of the very first concerns surrounding online symptom
checkers for self-diagnosis and triage.29 It is very likely that AI driven software may
make mistakes in patient diagnosis leading to undesirable outcomes and it may be
difficult to establish accountability for such mistakes 30. Symptom checkers may also
be subject to algorithmic bias, perhaps predicting greater likelihood of disease on the
basis of gender or race when those are not actually causal factors 31. Mechanisms, such
as audits to ensure accountability and responsibility for AI systems and their output
should be put in place32.

Transparency is vital to ensure that AI is not biased33. The EU AI guidelines introduce

several measures to guarantee transparency in the AI industry34. For example, AI
systems, like online symptom checkers should be identifiable as such, and users need
to be aware that they are interacting with an AI algorithm and not with a health
practitioner35. Informing the users of the peculiarities and operating of these
algorithms involved in the self-diagnosis and triage will be crucial, in order to
understand that there is room for error as well as for determining liability.

Innovations in health care sector affect relationships between patients and doctors.
With diagnosis provided remotely by an algorithm, human touch and other forms of

28
Article 19 (2018).Privacy and Freedom of Expression in the Age of Artificial Intelligence.
https://www.article19.org/wp-content/uploads/2018/04/Privacy-and-Freedom-of-Expression-In-the-
Age-of-Artificial-Intelligence-1.pdf (Last Access: April 20, 2020)
29
Davenport, T., & Kalakota, R. (2019). The potential for artificial intelligence in healthcare. Future
healthcare journal, 6(2), 94–98. https://doi.org/10.7861/futurehosp.6-2-94. (Last Access: April 20,
2020)
30
Ibid.
31
Ibid.
32
McKinsey & Company (2020). Transforming healthcare with AI. The impact on the workforce and
organisations. https://ec.europa.eu/digital-single-market/en/news/ethics-guidelines-trustworthy-ai (Last
Access: April 20, 2020)
33
Tambiama Madiega (2019)EU guidelines on ethics in artificial intelligence: Context and
implementation. European Parliament.
https://www.europarl.europa.eu/RegData/etudes/BRIE/2019/640163/EPRS_BRI(2019)640163_EN.pdf
(Last Access: April 28, 2020)
34
Ibid.
35
Ibid.

7
 Online symptom checkers for self-diagnosis and triage

verbal communication are lacking, leading from a personal to an impersonal

relationship with an invisible “robo-doctor”.

Regarding the collections of sensitive information gathered on online symptom

checker platforms or applications, some assurances should set for in order to ensure
that the information cannot be used maliciously if accessed or intercepted without
authorization. From a legal standpoint, anonymization and pseudonymization are
acceptable measures, as GDPR mechanisms for preventing personal data exposure.
Another crucial prerequisite is to have secure and reliable systems and software.
Trustworthy AI calls for secure, reliable, and robust algorithms that are able to handle
errors or inconsistencies in all phases of an AI system's life cycle 36. This prerequisite
is about ensuring cybersecurity37. In practice, all vulnerabilities should be considered
when developing algorithms38. This includes checking AI systems to identify and
reduce the cyber-attacks and hacking risks39. Attacks may target the data (data
poisoning), the model (model leakage), or the underlying infrastructure, both software
and hardware40. If an AI system is attacked, the data as well as algorithm behavior
may change, leading the system to make various erroneous decisions 41. AI developers
will incorporate procedures that can determine the safety risks involved in the event
that anyone uses the AI system they are developing for harmful purposes42.

Another dilemma concerns equity in access to health information. There is a doubt

whether online health checkers are a useful solution for developing countries and poor
people or those who live in remote areas or whether, instead, new technological
developments will benefit only those able to afford them, since an Internet connection
is definitely required, sometimes along with a subscription fee for more sophisticated
diagnosis.

4. Socio-political Consequences of using Online Symptom Checkers

Like every new medical technology, online symptom checkers for self-diagnosis and
triage aim at the well-being of the society since more people can be informed for their
health condition and be treated. The regulation and the use of online symptom
checkers could encourage innovation and the better use of technology. The
36
Tambiama Madiega (2019)EU guidelines on ethics in artificial intelligence: Context and
implementation. European Parliament.
https://www.europarl.europa.eu/RegData/etudes/BRIE/2019/640163/EPRS_BRI(2019)640163_EN.pdf
(Last Access: April 28, 2020)
37
Ibid.
38
Ibid.
39
Ibid.
40
High-Level Expert Group on Artificial Intelligence (2019). Ethics Guidelines for Trustworthy AI.
European Commission. https://ec.europa.eu/digital-single-market/en/news/ethics-guidelines-
trustworthy-ai (Last Access: April 28, 2020)
41
Ibid.
42
Tambiama Madiega (2019)EU guidelines on ethics in artificial intelligence: Context and
implementation. European Parliament.
https://www.europarl.europa.eu/RegData/etudes/BRIE/2019/640163/EPRS_BRI(2019)640163_EN.pdf
(Last Access: April 28, 2020)

8
 Online symptom checkers for self-diagnosis and triage

manufacturers and the designers of the technology will be reassured that their
investments and innovative applications will not be futile, but they will be used and
will have an impact on the research and development of online symptom checkers for
self-diagnosis and triage, whereas technology will be used for good purposes.

Online symptom checkers for self-diagnosis and triage, would gradually redirect
patients to appropriate treatments for their particular symptoms and underlying
conditions43. This may result to the reduction of health care professionals’ workload,
especially in primary and emergency care, and frees them up to address patients with
urgent needs44.

Online symptom checkers could definitely lead to the creation of new professions
with specialization in this medicine sector. Several professions such as, data scientists,
AI engineers, data engineers, data governance experts etc. will arise from the
combination of medical and data-science expertise. Implementing AI solutions will
require new roles in training, remodelling workflows and communicating the benefits
of AI45. However, at the same time, online symptom checkers could cause the loss of
jobs in the medical and healthcare sector, by putting pressure on the “traditional”
health care professionals, since patients might prefer asking their online doctor about
their health condition rather than go to a real doctor.

The application of online symptom checkers might urge reactions within the society
and the political parties, between those in favor and against the usage of online
symptom checkers, with a party to stress the unavoidable involvement of the
providers of online symptom checkers to a large extent in healthcare. A fact that may
hamper or slow down the development of the technology.

5. Socio-political Consequences of Different Regulatory Choices

5.1. No Regulation
As stated, before online symptom checkers are already in use, thus not regulating their
usage is not an option, since issues may arise against people who use such
applications. Legal framework on EU level is important because no regulation can
lead to applicability of the national law of each member state. A fact that could lead to
discrepancies between member states, depending on the legal system and rules in
place.

5.2. Prohibit online symptom checkers

Prohibiting online symptom checkers is not effective solution, because this
technology has already been implemented to a large extent. Thus, the choice of

43
McKinsey & Company (2020). Transforming healthcare with AI. The impact on the workforce and
organisations. https://ec.europa.eu/digital-single-market/en/news/ethics-guidelines-trustworthy-ai (Last
Access: April 20, 2020)
44
Ibid.
45
Ibid.

9
 Online symptom checkers for self-diagnosis and triage

banning it is not an effective way of managing the situation. A ban of online symptom
checkers will discourage innovation and the better use of technology. Furthermore,
since online symptom checkers are in use , any investments already happened would
be in vain. Forbidding online symptom checkers in general may lead to the avoidance
of the problems that will arise from this new technology, but at the same time there
will be no solution to the oldest problems, as well as no opportunity to surpass the
new challenges. Furthermore, it could limit the economic discrepancies that may rise
in the initial stage, since every innovative technology is expected to be more
expensive in its early stages.

5.3. Regulating online symptom checkers without safeguards

Just permitting the usage of online symptom checkers without any specific regulation
would only enlarge the existing issues. Health care sector would be easily
manipulated by market actors, while political policies that will be established would
affect the social behaviors and attitude as a whole. Someone could allege that without
provision of safeguard, the society, the market, and the patients would have more
freedom of choice and they would be able to customize treatments upon to their
discretion, according to their needs.
 

6. Providers’ Contribution to governance and use of online symptom checkers

According to M. Huffy “governance comprises all of the processes of governing

whether undertaken by the government of a state, by a market or by a network over a
social system and whether through the laws, norms, power, or language of an
organized society. It relates to the processes of interaction and decision-making
among the actors involved in a collective problem that lead to the creation,
reinforcement, or reproduction of social norms and institutions46.”
In this context, providers could play an essential role in governance and the
use of this technology in society. Being one of the sectors (the market) that govern the
potential, the characteristics and the application of online symptom checkers, through
the power they acquire, as the manufacturers and the distributors of this technology
and the required equipment, respectively, their contribution in governance is major.
Providers are responsible for shaping the trajectory of the technology for positive
outcomes. Implementing AI powered online symptom checkers efficiently will
require re-skilling the workforce to effectively embrace new technologies 47. To ensure
success, it will be important for providers to foster these capabilities in the

46
Hufty, Marc, Investigating Policy Processes: The Governance Analytical Framework (GAF) (2011).
RESEARCH FOR SUSTAINABLE DEVELOPMENT: FOUNDATIONS, EXPERIENCES, AND
PERSPECTIVES, pp. 403-424, U. Wiesmann, H. Hurni, ed., Geographica Bernensia, 2011. Available
at SSRN: https://ssrn.com/abstract=2019005 (Last Access: May 20, 2020)
47
World Economic Forum (2017). Technology and Innovation for the Future of Production:
Accelerating Value Creation. White paper. Available at:
http://www3.weforum.org/docs/WEF_White_Paper_Technology_Innovation_Future_of_Production_2
017.pdf (Last Access: April 7,2020)

10
 Online symptom checkers for self-diagnosis and triage

workforces, through training, retraining, and offer internships and short courses on AI
technology in medicine48.
Precision concerns the ability of an AI online symptom checker to make
accurate judgments, such as correctly classifying information into the relevant
categories, or its capability to make correct recommendations and decisions based on
the input data49. A clear and well-formed process of development and assessment will
help, minimize, and correct unintended risks from imprecise predictions50.
The quality of the data sets used is vital to the performance of AI systems 51.
When collecting data, it can contain inaccuracies, biases, mistakes, and errors 52.
Feeding erroneous data into an AI software, such as the online symptom checkers,
might change its behavior, especially with self-learning systems. Data and procedures
must be checked and documented continuously. The way in which AI systems are
developed may also suffer from unfair bias, which could lead to unfair decisions and
discrimination53. So, developers of online symptom checker softwares should make
sure that their products are free from biases which could lead to unfair diagnosis and
triage.

7. Conclusion

AI, in general, is a promising new technology that brings much hope and has brought
a lot of hype in the past54. The future is full of possibilities, which may lead to
intended and unintended consequences in a variety of domains55.

AI and automation may potentially change the way healthcare is provided, addressing
both the need for quality and more cost-effective treatment, while helping to fill some
of the staff shortage56. Speeding up the pace, through the careful and systematic
implementation of AI, will bring major benefits for patients and communities across
Europe57. Europe should draw on the exceptional strengths of its national health
systems and datasets and its innovation ecosystem to make sure that the rights of

48
World Economic Forum (2020). 3D printing: A Guide for Decision-Makers. White paper. Available
at: http://www3.weforum.org/docs/WEF_Impacts_3D_Printing_on_Trade_Supply_Chains_Toolkit.pdf
(Last Access: April 7,2020)
49
High-Level Expert Group on Artificial Intelligence (2019). Ethics Guidelines for Trustworthy AI.
European Commission. https://ec.europa.eu/digital-single-market/en/news/ethics-guidelines-
trustworthy-ai (Last Access: April 28, 2020)
50
Ibid.
51
Ibid.
52
Ibid.
53
Ibid.
54
Ibid
55
Ibid
56
McKinsey & Company (2020). Transforming healthcare with AI. The impact on the workforce and
organisations. https://ec.europa.eu/digital-single-market/en/news/ethics-guidelines-trustworthy-ai (Last
Access: April 20, 2020)
57
Ibid.

11
 Online symptom checkers for self-diagnosis and triage

patients to their data remain inviolable whilst guaranteeing that the same patients
benefit from the enormous AI promise in healthcare sector58 .

Like AI powered online symptom checkers, every new technology, jeopardize the
balance of the traditional regulatory health structure. This fact calls for regulatory
change, a process where painstaking efforts are required, in order the interests of all
actors be served, while primarily fostering the right to health and the safest use of the
technology at stake. These efforts should be considered as an integral part of EU’s
health policy dialogue involving stakeholders from health, civil society, finance,
professional associations, international agencies, nongovernmental organizations,
academic institutions, and communities59.

At the moment, responsibility for AI adoption and its impact in the healthcare sector,
sits with healthcare practitioners or healthcare organizations based on national
liability regimes60. This creates insecurity and delays the adoption of AI driven
solution in healthcare. Providers need assurances that their AI applications are
authorized for use and certainty on who bears the risk.

Law as one of the modalities governing regulation should undoubtedly play an

essential role when regulating a new technology.

In this respect, policy-makers can prepare themselves for the new challenges derived
from AI in medicine by: (i) developing European strategies for promoting safety in AI
medical procedures; (ii) adjusting existing regulations/laws and/or developing new
ones to prepare for potential unintended consequences; and (iii) incorporating leading
indicators in periodic planning processes for trade, customs, liability regimes and
cybersecurity, among other areas61.

58
Ibid.
59
Clarke D. Chapter 10. Law, regulation and strategizing for health. In: Schmets G, Rajan D,
Kadandale S, editors. Strategizing national health in the 21st century: a handbook. Geneva: World
Health Organization; 2016. Available at: https://www.who.int/healthsystems/publications/nhpsp-
handbook-ch10/en/ (Last Access: April 10,2020)
60
McKinsey & Company (2020). Transforming healthcare with AI. The impact on the workforce and
organisations. https://ec.europa.eu/digital-single-market/en/news/ethics-guidelines-trustworthy-ai (Last
Access: April 20, 2020)
61
Ibid

12