Open Access Research Article
CHALLENGES OF ARTIFICIAL INTELLIGENCE IN HEALTH CARE SYSTEM: ITS LEGAL AND ETHICAL CONSEQUENCES.
Published Paper
Article Details
CHALLENGES OF ARTIFICIAL INTELLIGENCE IN HEALTH CARE SYSTEM: ITS
LEGAL AND ETHICAL CONSEQUENCES.
AUTHOR- HRISHIKESH RAJKHOWA
Abstract-
A sizable set of methods for treating diseases have been
developed in medical practice. The high investment allure of medical
technologies seems to be leading hospitals to do actions that are against the
Hippocratic oath of doctors. The widespread employment of cyborg-AI doctors, AI
robots, and AI medical organizations in the near future may be hampered by
this. Of course, choosing the best course of treatment requires having the
accurate diagnosis. The AI-robot recently exhibited the capacity to choose appropriate
treatment procedures and efficient medications based on genomic data. However,
this study is not for the discussion of the various advantages that AI has and
ratter it focuses on the problems that are upcoming because of the introduction
of AI in the sector of health care and medical science. The problems ranges
from various aspects such as patients safety, doctor and patient relationship,
transparency, accountability, privacy etc. This study makes a general overview
of the existing use and problems of AI in health care and at the same time
trace any law that stands existing in the world for its regulation.
Keywords- AI-
Artificial intelligence, IBM- International Business Machines, ML- Machine
Learning, GDPR- General Data Protection Regulation, CDS- Clinical Decision
Support, EU- European Union,
I.
INTRODUCTION
The term "artificial
intelligence" (AI) refers to a group of technological solutions that mimic
human cognitive functions, including the capacity for independent learning and decision-making
without the aid of predetermined algorithms. When used to accomplish certain
tasks, AI can yield results that are on par with or superior to those produced
by human intellectual activity. This group of technological solutions consists
of tools and services for data processing, decision-making, and information and
communication infrastructure, software (including program that use machine
learning techniques), and software applications.
The evolution of how digital
technology and artificial intelligence (AI) have been used in healthcare has
evolved over many years. MYCIN, an expert system developed by Stanford
University enabled medical professionals to recognize bacterial diseases such
bacteremia and meningitis, and to recommend a suitable course of action.
MYCIN was not utilized in daily life. It was just used as an experimental model
to show what AI is capable of. In 1986, with the help of a decision
assistance system called DXplain, the University of Massachusetts using
the patient's symptoms created a list of potential diagnoses which
was generated for the doctor's reference. The University of
Washington then put the Germwatcher expert system into practice for the
patient's identification of infections[1].
Since the start of the creation of AI-based medical applications in the
twenty-first century has been a priority concern for IT based industries.
According to IBM, the company that
created the Watson supercomputer, AI technologies can be used in healthcare to
structure medical data. For example- processing natural language and
turning it into clinical text, analyze patient information, such
as abstracting treatment records into a patient's medical history
comparing clinical diagnostic findings to choose the best course of
action, advancing medical philosophy by developing models of patient
therapy based on comparable examples, as well as confirming medical theories.[2]
II. USE OF ARTIFICIAL INTELLIGENCE IN
HEALTH CARE AND MEDICINE.
The various sectors where AI has been
extensively used in the field of AI and health care are as follows.
1. Development of Drug and its
Validation-
In the process of developing new
medications, AI first made it possible to speed up the work using massive data.
In this context, the fundamental goal of AI is to forecast how future drug
molecules will interact with the proteins of human cells and, in turn, how well
the medicine will work. AI can also be used to investigate illness mechanisms,
find biomarkers, and study them.
In order to quickly create a
treatment for Ebola virus infections in 2015, Atomwise collaborated with the
University of Toronto and IBM amid an outbreak of the disease in West Africa.
Atomwise has made fundamental AI technology available for medicinal
development.[3]
For instance, AI technology allowed
researchers to examine the activity of dozens of medications in 2020 in
connection to their capacity to inhibit an enzyme that the SARS-CoV-2 virus
needs to replicate in human cells. The application of AI technology has
enhanced the design of clinical trials, streamlined the development of novel
vaccinations and the analysis of trial findings, in addition to the comparison
and systematization of data from various patient groups in the perspective of
the COVID-19 pandemic.[4]
2. Diagnosis of disease and off-site AI
application-
The use of algorithms in medical AI
offers many advantages over human cognition, such as the capacity to work
nonstop and the absence of vulnerability to weariness or emotional bias. In the
event of a disease outbreak (epidemics and pandemics), the treatment of severe
types of diseases, and the appearance of new diseases that were previously
unknown to medicine and such other advantages become extremely crucial.
Physicians attest to AI's capacity to correctly identify a variety of
illnesses. For example it has been shown that AI can accurately identify colon
polyps during a colonoscopy and can identify coagulopathy and inflammation in
trauma.[5] AI
is able to evaluate a patient's present state of health and instantaneously
place it in the context of their whole medical history. One use of this
functionality is in Botkin which is a Russian software that detect
cancer and provide answers in 2019.
Additionally, AI can assess external
factors such as weather and climatic conditions (temperature, atmospheric
pressure, and humidity level), sanitary and epidemiological conditions, a
patient's genetic susceptibility to certain infections, economic factors
(household income, living arrangements, and working capacity), degree of
socioeconomic status, and environmental factors in real time. All forms of AI,
including a cyborg-AI doctor, a robot doctor, an AI hospital, and an AI cloud
doctor, can be used to address the aforementioned problems. AI can help in
managing health issues, preventing diseases, and keeping an eye on the
likelihood that diseases will spread.[6]
A remote medical examination of a
patient based on their symptoms and medical history, and the determination of
the need for hospitalization (based, for example, on the results of an ECG,
coronary angiogram, or ultrasound examination) are all examples of ways that AI
can be used in healthcare off-site. These areas include oncology,
gastroenterology, orthopedics, ophthalmology, endocrinology, gynecology, and
others.
Obviously, a home X-ray, MRI, or CAT
scan are not now available (off-site). In many situations, having physical
contact with a patient is still necessary. Only a few medical tests may be
performed at a patient's house because to the portable equipment that is now
available. Therefore, the deployment of cyborg-AI-doctors, AI-robots, and
AI-medical organizations in healthcare facilities is required. Although an
AI-cloud doctor is not yet able to conduct a thorough patient examination from
a distance, some encouraging progress has been made in this area.[7]
3. Treatment by noble AI powered
solution.
The field of medicine has
produced a sizable set of methods for treating diseases. It appears that
hospitals may make decisions that are against the Hippocratic oath of doctors
due to the high investment attraction of healthcare technology. This could
limit the widespread usage of AI-cloud-doctors, as well as the predominance of
cyborg-AI-doctors, AI-robots, and AI-medical organizations in the near future.
It should be obvious saying that selecting the proper treatment requires having
the accurate diagnosis. The AI robot Sophia recently exhibited the capacity to
choose appropriate treatment procedures and efficient medications based on
genomic data[8].
All types of AI,
including cyborg-AI-doctors, AI-robots, AI-hospitals, and AI-cloud-doctors, can
be used effectively in the following fields:
·
medical
interventions (surgery), pharmaceutical production and prescription
(pharmaceutics and pharmacology),
·
immunotherapy and herbal medicine, epidemic
epidemiology, etc.
·
A
cyborg AI doctor
·
AI-robot/AI-hospital/AI-cloud
doctor supporting a human physician, or
·
an
autonomous and remotely controlled AI-robot/AI-hospital are the three various
ways AI can function.
The chances of interacting
with various types of AI appear good. An AI-robot assistance, for instance,
could be useful to a human surgeon performing an intervention using a
remote-controlled machine with surgical instruments. AI can quickly access a
patient's medical history during both surgical and diagnostic procedures and
assess the variables that may influence the choice of treatment (climatic
circumstances, epidemiological situation, the patient's genetic susceptibility
to infections, etc.).
Robotic surgery and
pharmaceuticals, which would enable lowering the expenses of staffing and
24-hour patient care, are the future of medicine. Even a medication driven by
mediocre AI has the potential to help a patient by inducing the placebo effect.
Furthermore, due to the fact that AI's algorithms are created to minimize
errors, patients will be more inclined to believe that the work of AI is
error-free on a subconscious level.
III
CHALLENGES IN THE USE OF ARTIFICIAL INTELLIGENCE IN HEALTH CARE AND MEDICINE.
The application of AI in clinical healthcare practice has enormous
potential to improve it, as the previous section suggested, but it also poses
many ethical and legal issues when used in this sector that are necessary to be
examined.
1.
Consent for use-
The patient-physician interaction will change
as a result of health AI applications in areas like imaging, diagnosis, and
surgery. But how will the application of AI to health care interact with the
concepts of informed consent? Although informed consent will be one of the most
urgent obstacles to successfully Integrating into clinical practice, this
burning question has not earned enough attention in the ethical discussion. It
is necessary to investigate when the informed consent principles should be used
in the context of clinical AI.[9]
How much of a duty does clinicians
have to inform patients on the intricacies of AI, especially the type of ML
used? The data inputs used, the system, and the potential for biases
or are there any other flaws in the data being used. In what
conditions must the patient be informed that artificial intelligence is
being employed at all?
These inquiries are particularly
difficult to respond to when the AI uses "black-box" algorithms,
which may be the consequence of imprecise machine-learning methods that are
highly complex for therapists to completely comprehend. For instance, Corti's
algorithms[10] are
"black boxes" since not even the software's creator is aware of how
the program determines when to notify emergency dispatchers that a person is
having a cardiac arrest. Medical practitioners might be concerned about this
lack of understanding. How much should a clinician, for instance, explain that
they are unable to completely understand the AI's diagnostic or therapy
suggestions? How much openness is required? How does this interact with the
GDPR's purported "right to explanation"? What about situations when
the patient would be hesitant to consent to the use of specific data categories
(such genetic information and family history)? How can we effectively strike a
balance between patient privacy and the security and efficiency of AI?
AI healthcare apps and chatbots are
also being utilized more often in a variety of health-related applications,
such as diet advice, health evaluations, assistance with medication adherence,
and analysis of data gathered by wearable sensors. These applications
raise concerns from bioethicists regarding user agreements and how they
relate to conscious consent as opposed to the conventional. A user agreement is
a contract that a person signs following the informed consent
process without a face-to-face conversation. Most individuals habitually
disregard user contracts because they don't spend the time to study them.[11] People's
ability to adhere to the terms of service they have accepted is further
complicated by the software's frequent updates. What details should be provided
to users of these apps and chatbots? Do users fully comprehend that continuing
to use the AI health app or chatbot may require agreeing to modified terms of
use? How much should informed consent documents mirror user agreements?[12]
2.
Safety and Transparency-
One of the main obstacles for AI in healthcare
is safety. To give an example that has received a lot of attention, IBM Watson
for Oncology employs AI algorithms to analyse data from patient medical records
and assist doctors in looking into cancer therapy options for their patients.
It has recently drawn criticism, meanwhile, for allegedly making "unsafe
and erroneous" suggestions regarding cancer treatments. The issue appears
to be with Watson for Oncology's training process, which only used a small
number of "synthetic" cancer cases developed by Memorial Sloan
Kettering (MSK) Cancer Center physicians as opposed to genuine patient data.[13]
The field has received negative
publicity as a result of this actual instance. It also demonstrates how crucial
it is that AIs be reliable and efficient. But how can we make sure
that AIs maintain their word? Stakeholders, in particular AI developers,
must ensure two critical factors in order to fully utilize AI's potential: (1)
the authenticity and trustworthiness of the datasets; and (2) transparency.
The employed datasets must first be
valid and dependable. When it comes to AI, the saying "garbage in, trash
out" is applicable. The performance of the AI will improve with greater
training data or the labelled data. To get reliable results, the
algorithms frequently need to be improved further. Data sharing is a
significant problem as well. For self-driving cars, for example, where the AI
needs to have a high level of confidence, more data sharing will be required.[14]
Second, some degree of transparency must be
guaranteed in the interest of patient confidence and safety. While in an ideal
world all information and algorithms would be accessible to the public, there
may be some real concerns about safeguarding investments and intellectual
property as well as avoiding an increase in cybersecurity risk. Additionally,
AI developers must be sufficiently transparent, for instance, regarding the
types of data used and any software flaws such as data bias. We should draw
conclusions from cases like Watson for Oncology, in which IBM concealed
Watson's risky and ineffective therapy suggestions for more than a year.[15] Finally,
trust among all parties, especially physicians and patients, is built through
transparency, and this trust is essential for a successful application of AI in
clinical practice.[16]
3.
Fairness and Algorithm
Biasness.
AI has the potential to enhance
healthcare not just in high-income settings, but also by
"globalizing" it and making it accessible to even remote locations.
Any ML system or human-trained algorithm, however, will only be as reliable,
efficient, and equitable as the data it is taught with. AI is also susceptible
to biases, which could lead to prejudice.[17]
Therefore, it is crucial that AI developers are conscious of this danger and
take steps to reduce any potential biases at every phase of the product
development process. When choosing (1) the ML procedures they want to employ to
train the algorithms and (2) what datasets including evaluating their quality
and diversity, they want to utilize for the programming, they should pay
particular attention to the risk for biases.
Algorithms can display biases that
can lead to unfairness with regard to ethnic origins, skin colour, or gender,
as shown by a number of real-world incidents. Biases might also exist with
reference to different characteristics, including age or a disability.[18] Such
prejudices have numerous, varied, and complex justifications. For instance, they
might be the outcome of the datasets themselves (which are not representative),
the selection and analysis of the data by data scientists and ML systems, the
use of AI, etc.
Biased AI could, for example, result
in incorrect diagnoses, render therapies ineffective for specific subgroups,
endanger their safety, and so forth in the medical sector when phenotype- and
occasionally genotype-related information is involved. Consider an AI-based
clinical decision support (CDS) tool that enables doctors to help patients with
skin cancer receive the optimal care. The algorithm, however, was primarily
trained on patients who were Caucasian. As a result, the AI software will
probably provide less accurate or even incorrect suggestions for subpopulations
like African Americans for which the training data was underinclusive.
4.
Data Privacy-
The Royal Free NHS Foundation Trust
was found to have violated the UK Data Protection Act 1998 when it gave Google
DeepMind access to the personal information of about 1.6 million patients,
according to a decision made in July 2017 by the UK Information Commissioner's
Office (ICO). Data exchange took place for "Streams", an app
that intends to assist with the diagnosis and identification of acute kidney
injury, throughout the clinical safety testing phase. Patients weren't properly
informed about how their data was processed during the exam, though. Elizabeth
Denham of the Information Commissioner properly noted that "the erosion of
fundamental privacy rights does not need to be the price of innovation."[19]
Recent case studies demonstrating
patient privacy issues in the context of data sharing and the application of AI
include the legal proceeding Dinerstein v. Google[20] and
Project Nightingale by Google and Ascension[21].
What about who owns the data, though? Health data can be worth billions of
dollars, but some data indicates that the public is uneasy about businesses or
the government making money off of selling patient data[22].
However, there can be other means by which patients can feel valued than actual
ownership. However, there can be other means by which patients can feel valued
than actual ownership. Enabling instance, the Royal Free NHS Foundation Trust
and Google DeepMind reached an agreement for the Trust to use Streams for free for
the next five years in exchange for providing patient data for the app's
testing. Ownership is not always necessary for reciprocity, but anyone wishing
to utilize patient data must demonstrate how doing so will benefit the health
of the very same patients whose data is being used.
IV.
LAW REGULATING AI
IN THE FIELD OF HEALTH CARE AND MEDICINE.
The chances of
interacting with various types of AI appear good. For instance, a human surgeon
doing an operation while using a remote-controlled device, a n AI-robot
assistance may be useful for those using surgical equipment. lowering the
cost of hiring workers and providing 24/7 patient care. Even inadequate or
subpar AI-driven treatment may be advantageous by inducing the placebo
effect in a person. Additionally, patients will unconsciously be more prone to
view the AI's work is error-free since its algorithms are created to
reduce errors.[23]
A. Legal Aspects of AI
in the field of medicine and health care-
AI and the issue of
trust in the field of Medical Science. The development of a single, trustworthy
digital realm has been made possible by advances in science, technology, and
digital technology. Participants in this area are assumed to have confidence in
the information obtained from it, and as a result, their identification and
authentication take place automatically. Regarding the acceptance of electronic
signatures, the topic of the "space of trust" was initially raised.
The European Union passed Directive 1999/93/EC of the European Parliament and
of the Council of December 13, 1999, on a Community Framework for Electronic
Signatures to address this problem. This directive was later replaced by
Regulation (EU) No 910/2014 of the European Parliament and of the Council of
July 23, 2014, on Electronic Identification and Trust Services for Electronic
Transactions in the Internal Market and Repealing Directive 1999/93/EC.[24]
The Eurasian Economic
Union and other international organizations used a similar strategy.
Information could be electronically transmitted remotely thanks to the
Internet. Through the identification and verification of participants in the
information exchange, the existence of the digital space of trust with regard
to electronic signatures ensures that the information received is trustworthy
and reliable.[25]
In general, any digital
information is decoded by translating binary code—a series of ones and
zeros—into text that can be understood by humans. It seems sensible to utilize the
using the e-signature space of trust as a guide for creating a digital space of
trust in the AI, where uniform identification and permission procedures would
apply to medical AI. The creation of a single digital platform for AI trust is
anticipated to be hindered by various procedures implemented in medical schools
and medical customs around the world (the subjective factor). However, all
medical professionals share a common bond regardless of ethnicity, to treat the
patient or prevent them from becoming ill religion, gender, and color, as well
as the hospital's location. As a result, it is likely that international
medical alliances and organizations (such as the World Health Organization, the
International Federation of Red Cross and Red Crescent Societies, Médecins Sans
Frontières, etc.) will play a significant part in the development of the AI
trust space in the digital sphere.
Doctors, patients, the
government, the state, and civil society as a whole will all recognize the
veracity of the information transmitted related to AI, which is the fundamental
tenet guiding the creation of the unified digital space of trust in medical AI.
B. Laws in European Union.
The following sources of legal
regulation apply at the level of the European Union (EU): The application of AI
technology can be identified.
· The EU Parliament passed Resolution
2015/2103(INL) Civil Law in February 2017. Robotics regulations.[26] Robotics
regulations in Europe AI is founded on Isaac Asimov's rules, which state that:
(1) A robot shall not, through act or omission, cause
(2) A robot must accept human
commands if they do not conflict with the first in order to avoid harm to
humans.
(3) A robot is required to ensure its
safety to the degree that doing so does not violate the one of the two laws.
The terms of the aforementioned
regulation primarily apply to robotics, although it might be assumed that, by
analogy, they also apply to AI technology. The resolution proposes giving AI
robots the status of electronic persons, establishing a European registration
system for "smart" robots, and defining culpability for harm caused
by robotics.
Following this, representatives of 25
European nations, including those outside the European Union, signed a
Declaration of cooperation on artificial intelligence in April 2018. According
to its provisions, the signatory nations committed to working on an integrated
European approach to the development of artificial intelligence, pursuing
cogent national policies to increase the competitiveness of the European Union,
and fostering digital innovation.[27]
The Coordinated Plan for Artificial Intelligence
of December 7, 2018, which offers a European strategy for the development of
robots and AI, was created by the European Commission also in 2018. The
participating States' overarching objective in cooperating is to make sure that
Europe emerges as the world's premier region for the development and
implementation of more advanced, ethical and safe AI.[28]
Additionally, ethical concerns
related to the usage of AI are regulated. In order to promote trustworthy AI,
the European Commission approved Ethics Guidelines for Trustworthy AI in 2019
(Ethics Guidelines for Trustworthy AI. High-Level Expert Group on Artificial
Intelligence 2019). Three requirements for trustworthy AI should be met over
the course of the system's complete life cycle: In order to ensure adherence to
ethical principles and values, it should be (1) lawful,
(2) ethical, and
(3) resilient from both a technical
and social standpoint. Even with the best of intentions, AI systems have the
potential to hurt people unintentionally. The
contents of this document state that respecting human autonomy, preventing
harm, being fair, and being explicable are the four basic ethical
considerations for using AI.
In order to digitalize European society and
preserve competitiveness in cutting-edge technologies, such as robotics and AI,
the EU has also adopted the Digital Europe Strategy Programme as a structural
component of the EU financial consolidation for 2021–2027.[29] This program's main goal is
to promote digital transformation by funding the adoption of emerging
technologies in the most crucial fields, each with a separate budget, such as
high performance computing, artificial intelligence, cyber security, and trust,
as well as the implementation and best use of digital literacy.
C. Laws in Russia
Russia
has been creating the initial iterations of national guidelines in the area of
AI in healthcare since 2020. The Technical Committee for Standardization
"Artificial Intelligence" (TC 164), founded on the model of the
Russian Venture Company, will oversee the development. Thus, by 2027, it is
intended to develop about 50 standards pertaining to the use of AI technology
in healthcare in specific areas, such as general requirements and
classification of AI systems in clinical medicine, radiology and functional
diagnostics, remote monitoring systems, histology, medical decision support
systems, image reconstruction in diagnostics and treatment, big data in
healthcare, medical analytics, and forecasting.
A
distinct type of law (rules of conduct) can be recognized in addition to the
conventional legal sources—a medical custom.
René
David and Camille Jauffret-Spinosi, researchers in comparative law, contend
that although legal customs need to be given legal legitimacy, this does not
preclude them from being regarded as an independent, impartial source of law.[30]
Legal traditions are typically accepted by academics as sources of law on par
with legislative acts (for examples, see Panagiotis Zepos' writings). The works
of Raymond Legeais provide a complete description of the contribution of legal
customs to the system of sources of law in various legal traditions. In line
with this theory, it might be claimed that medical legal traditions recognised
by the government or other public organizations (such as the World Health
Organization) can be seen as sources of AI control.
The
topic is not without debate, though. How do we combine medical conventions,
which have been formed over many years of effort by the medical community
(cyborg-AI-doctor, AI-robot, AI-medical organization, and AI-cloud-doctor),
into the legal framework governing AI technologies? Can such legal practices
evolve in the future as a result of the cooperation between doctors and AI
systems? The short history of AI makes it challenging to provide solutions to
these queries. It is anticipated that the state will play a significant part in
these issues since its competent authorities will have to decide which medical
legal conventions (written or unwritten) may regulate AI in healthcare. The
distinction between legal and illegal medical customs, the latter of which is
not approved by the state and does not serve as a source of legislation, is
crucial.
It is
necessary to standardize and make AI accessible to medical legal customs. To
guarantee the caliber and objectivity of AI, this is essential. The rules and
procedures for using AI technology (such as an AI-hospital, AI-robot, and
cyborg-AI-doctor) will be heavily influenced by the legal system in a
particular nation.
We
will then need to develop a worldwide database of medical legal norms adopted
by all nations taking part in the integration effort, which will serve as the
foundation for global AI (i.e., for an AI-cloud-doctor).[31]
V.
CONCLUSION
Cybersecurity,
informed consent, and high standards of data protection high standards of
safety and effectiveness, algorithmic fairness, resilience and cybersecurity,
appropriate transparency and regulatory oversight, all of these crucial
elements must be taken into consideration and handled in order to successfully
develop an AI-driven healthcare system. In this sense, we must do more than
simply update the existing regulatory frameworks to reflect new technological
advancements. However, it is equally crucial to have political and public
debates that centre on the morality of AI-driven healthcare and its effects on
the workforce and society at large. AI has enormous potential to enhance our
healthcare system, but we can only fully realise that potential if we start now
to address the moral and legal issues we face.
[1]G. Michael, A. Steib, C.D. Wiliam, and F.J. Victoria,
“Monitoring expert system performance using continuous user feedback” 3 Journal
of the American Medical Informatics Association 216 (1996).
[2] What is Artificial Intelligence in Medicine? 2021, available
at <https://www.ibm.com/watson
health/learn/artificial-intelligencemedicine> (last visited on April 20th 2021)
[3] New Ebola Treatment Using Artificial Intelligence, available
at <https://www.atomwise.com/2015/03/24/new-ebolatreatment- using-artificial-intelligence/> (last visited on
22nd October 2021).
[4] Phulan Sarma, S. V. Rana, Bikash Medhi, and Manisha
Naithani, “Emerging role of artificial intelligence in therapeutics for
COVID-19: A systematic review” 10 Journal
of Biomolecular Structure and Dynamics 16 (2020).
[5] A. Sami, “Challenges facing
the detection of colonic polyps: What can deep learning do?” 55 Medicina
Journal 473 (2019).
[6] Wenbo Yang, Jehane Michael Le Grange,
Peng Wang, Wei Huang, and Ye Zhewei,
“Smart healthcare: Making medical care more intelligent” 3 Global
Health Journal 65 (2020).
[7] Eric Campo, Daniel Esteve, and Jean-Yves
Fourniols. “Smart homes—Current features and future perspective” 64 Maturitas
97 (2019).
[8] Sophia AI Reaches Key Milestone by Helping to Better
Diagnose 200,000 Patients Worldwide, available at
(last visited on 22 May 2021).
[9] I.G Cohen, R. Amarasingham, A. Shah, B. Xie, “The
legal and ethical concerns that arise from using complex predictive analytics
in health care” 14 Health Aff
1134, available at <https://doi.org/10.1377/hlthaff.2014.0048.> (last visited on 5th 2021).
[10] J. Vincent, “AI that detects cardiac arrests during
emergency calls will be tested across Europe this summer” Verge, available
at <https://www.theverge.com/2018/4/25/17278994/aicardiac- arrest-corti-emergency-call-response 2018> (last
visited March 21st 2021).
[11] I.G, Cohen, A. Pearlman, “Smart pills can transmit
data to your doctors, but what about privacy?” N Scientist, May 2019, available
at <https://www.newscientist.com/article/2180158-smartpills-can-transmit-data-to-your-doctors-but-what-about-privacy>
(last visited on May 16th 2022).
[12] Ibid.
[13] C. Ross, I. Swetlitz, “IBM’s Watson supercomputer
recommended ‘unsafe and incorrect’ cancer treatments, internal documents show”,
STAT, Jan 2017, available at <https://www.statnews. com/2018/07/25/ibm-watson-recommended-unsafe-incorrect-treatments>
(last visited on 7th October 2021).
[14] Figure Eight. What is training data?, available at
<https://www.figure-eight.com/resources/whatis- training-data; 2020> (last visited on Jan 1st
2021).
[15] Ibid.
[16] J. Brown, “IBM Watson reportedly recommended cancer
treatments that were unsafe and incorrect”, Gizmodo, May 2019, available
at <https://gizmodo.com/ibm-watson-reportedly-recommended- cancer-treatments-tha-1827868882; 2018> (last
visited on Feb. 23rd 2021).
[17] B. Wahl, A. Cossy-Gantner, S. Germann, “Artificial
intelligence (AI)and global health: how can AI contribute to health in
resource-poor settings?”, BMJ Glob
Health, March 2018, available at (last visited on July 21st 2021).
[18] N. Sharkey, “The impact of gender and race bias in
AI”, Humanitarian Law Policy, Sept 2018, available at <https://blogs.icrc.org/law-and-policy/2018/08/28/impact-gender-race-bias-ai>
(last visited on Jan. 3rd 2021).
[19] ICO. Royal Free, “Google DeepMind trial failed to
comply with data protection law”, available
at <https://ico.org.uk/about-the-ico/news-and-events/news-and-blogs/2017/07/
royal-free-google-deepmind-trial-failed-to-comply-with-data-protection-law>
(last visited on Oct. 13, 2021).
[21] R. Copeland, “Google’s ‘Project Nightingale’ gathers
personal health data on millions of Americans”, Nightingalegathers, May
2018, available at <https://www.wsj.com/articles/google-s-secret-project-nightingalegathers-
personal-health-data-on-millions-of-americans-1157349679> (last visited on
March 23rd 2021)
[22] S. Gerke, I.G. Cohen, “Potential liability for
physicians using artificial intelligence”
JAMA, May 2019, available at <https://doi.org/ 10.1001/jama.2019.15064.> (last visited on July 21st
2021).
[23] Sophia AI Reaches Key Milestone by Helping to Better
Diagnose 200,000 Patients Worldwide, 2018. Available at <https://www.prnewswire.com/news-releases/sophia-ai-reaches-key-milestone-by-helping-to-better-diagnose-200000-patients-worldwide-680907791.html>
(last visited on March 21st 2021).
[24] Official Journal of the European Union, available
at <https://eur-lex.europa.eu/legal-content/EN/ TXT/PDF/?uri=OJ:L:2000:013:FULL&from=EN> (last
visited on May 15th 2021).
[25] Treaty on the Eurasian Economic Union, 2014, available
at https://www.un.org/en/ga/sixth/70/docs/treaty_on_eeu.pdf (last visited on 20th October 2021).
[26] European Parliament Resolution of 16 February 2017
with Recommendations to the Commission on Civil Law Rules on Robotics, 2015 available
at <http://www.europarl.europa.eu/doceo/document/TA-8-2017-0051_EN.pdf>
(last visited on Jan. 2nd 2021).
[27] EU Declaration of Cooperation on Artificial
Intelligence Signed at Digital Day on 10th April 2018, available at <https://ec.europa.
eu/digital-single-market/en/events/digital-day-2018> (last visited on May 6th
2021).
[28] Coordinated Plan on Artificial Intelligence 2021
Review, 2021, available at <https://digital-strategy.ec.
europa.eu/en/library/coordinated-plan-artificial-intelligence-2021-review>
(last visited on Nov. 23rd 2021).
[29] The Digital Europe Programme for the Period
2021–2027, 2018. Available at <https://eur-lex.europa.eu/legalcontent/ EN/ALL/?uri=CELEX:52018PC0434> (last visited on
Sept. 5th 2020).
[30] David, Rene, and Camille Jauffret Spinosi, Les
Grands Systèmes de Droit Contemporains 107 (The Major Contemporary Legal
Systems, Paris, 10th edn. 2002).
[31] Ibid.
AUTHOR- HRISHIKESH RAJKHOWA
Abstract-
A sizable set of methods for treating diseases have been
developed in medical practice. The high investment allure of medical
technologies seems to be leading hospitals to do actions that are against the
Hippocratic oath of doctors. The widespread employment of cyborg-AI doctors, AI
robots, and AI medical organizations in the near future may be hampered by
this. Of course, choosing the best course of treatment requires having the
accurate diagnosis. The AI-robot recently exhibited the capacity to choose appropriate
treatment procedures and efficient medications based on genomic data. However,
this study is not for the discussion of the various advantages that AI has and
ratter it focuses on the problems that are upcoming because of the introduction
of AI in the sector of health care and medical science. The problems ranges
from various aspects such as patients safety, doctor and patient relationship,
transparency, accountability, privacy etc. This study makes a general overview
of the existing use and problems of AI in health care and at the same time
trace any law that stands existing in the world for its regulation.
Keywords- AI-
Artificial intelligence, IBM- International Business Machines, ML- Machine
Learning, GDPR- General Data Protection Regulation, CDS- Clinical Decision
Support, EU- European Union,
I.
INTRODUCTION
The term "artificial
intelligence" (AI) refers to a group of technological solutions that mimic
human cognitive functions, including the capacity for independent learning and decision-making
without the aid of predetermined algorithms. When used to accomplish certain
tasks, AI can yield results that are on par with or superior to those produced
by human intellectual activity. This group of technological solutions consists
of tools and services for data processing, decision-making, and information and
communication infrastructure, software (including program that use machine
learning techniques), and software applications.
The evolution of how digital
technology and artificial intelligence (AI) have been used in healthcare has
evolved over many years. MYCIN, an expert system developed by Stanford
University enabled medical professionals to recognize bacterial diseases such
bacteremia and meningitis, and to recommend a suitable course of action.
MYCIN was not utilized in daily life. It was just used as an experimental model
to show what AI is capable of. In 1986, with the help of a decision
assistance system called DXplain, the University of Massachusetts using
the patient's symptoms created a list of potential diagnoses which
was generated for the doctor's reference. The University of
Washington then put the Germwatcher expert system into practice for the
patient's identification of infections[1].
Since the start of the creation of AI-based medical applications in the
twenty-first century has been a priority concern for IT based industries.
According to IBM, the company that
created the Watson supercomputer, AI technologies can be used in healthcare to
structure medical data. For example- processing natural language and
turning it into clinical text, analyze patient information, such
as abstracting treatment records into a patient's medical history
comparing clinical diagnostic findings to choose the best course of
action, advancing medical philosophy by developing models of patient
therapy based on comparable examples, as well as confirming medical theories.[2]
II. USE OF ARTIFICIAL INTELLIGENCE IN
HEALTH CARE AND MEDICINE.
The various sectors where AI has been
extensively used in the field of AI and health care are as follows.
1. Development of Drug and its
Validation-
In the process of developing new
medications, AI first made it possible to speed up the work using massive data.
In this context, the fundamental goal of AI is to forecast how future drug
molecules will interact with the proteins of human cells and, in turn, how well
the medicine will work. AI can also be used to investigate illness mechanisms,
find biomarkers, and study them.
In order to quickly create a
treatment for Ebola virus infections in 2015, Atomwise collaborated with the
University of Toronto and IBM amid an outbreak of the disease in West Africa.
Atomwise has made fundamental AI technology available for medicinal
development.[3]
For instance, AI technology allowed
researchers to examine the activity of dozens of medications in 2020 in
connection to their capacity to inhibit an enzyme that the SARS-CoV-2 virus
needs to replicate in human cells. The application of AI technology has
enhanced the design of clinical trials, streamlined the development of novel
vaccinations and the analysis of trial findings, in addition to the comparison
and systematization of data from various patient groups in the perspective of
the COVID-19 pandemic.[4]
2. Diagnosis of disease and off-site AI
application-
The use of algorithms in medical AI
offers many advantages over human cognition, such as the capacity to work
nonstop and the absence of vulnerability to weariness or emotional bias. In the
event of a disease outbreak (epidemics and pandemics), the treatment of severe
types of diseases, and the appearance of new diseases that were previously
unknown to medicine and such other advantages become extremely crucial.
Physicians attest to AI's capacity to correctly identify a variety of
illnesses. For example it has been shown that AI can accurately identify colon
polyps during a colonoscopy and can identify coagulopathy and inflammation in
trauma.[5] AI
is able to evaluate a patient's present state of health and instantaneously
place it in the context of their whole medical history. One use of this
functionality is in Botkin which is a Russian software that detect
cancer and provide answers in 2019.
Additionally, AI can assess external
factors such as weather and climatic conditions (temperature, atmospheric
pressure, and humidity level), sanitary and epidemiological conditions, a
patient's genetic susceptibility to certain infections, economic factors
(household income, living arrangements, and working capacity), degree of
socioeconomic status, and environmental factors in real time. All forms of AI,
including a cyborg-AI doctor, a robot doctor, an AI hospital, and an AI cloud
doctor, can be used to address the aforementioned problems. AI can help in
managing health issues, preventing diseases, and keeping an eye on the
likelihood that diseases will spread.[6]
A remote medical examination of a
patient based on their symptoms and medical history, and the determination of
the need for hospitalization (based, for example, on the results of an ECG,
coronary angiogram, or ultrasound examination) are all examples of ways that AI
can be used in healthcare off-site. These areas include oncology,
gastroenterology, orthopedics, ophthalmology, endocrinology, gynecology, and
others.
Obviously, a home X-ray, MRI, or CAT
scan are not now available (off-site). In many situations, having physical
contact with a patient is still necessary. Only a few medical tests may be
performed at a patient's house because to the portable equipment that is now
available. Therefore, the deployment of cyborg-AI-doctors, AI-robots, and
AI-medical organizations in healthcare facilities is required. Although an
AI-cloud doctor is not yet able to conduct a thorough patient examination from
a distance, some encouraging progress has been made in this area.[7]
3. Treatment by noble AI powered
solution.
The field of medicine has
produced a sizable set of methods for treating diseases. It appears that
hospitals may make decisions that are against the Hippocratic oath of doctors
due to the high investment attraction of healthcare technology. This could
limit the widespread usage of AI-cloud-doctors, as well as the predominance of
cyborg-AI-doctors, AI-robots, and AI-medical organizations in the near future.
It should be obvious saying that selecting the proper treatment requires having
the accurate diagnosis. The AI robot Sophia recently exhibited the capacity to
choose appropriate treatment procedures and efficient medications based on
genomic data[8].
All types of AI,
including cyborg-AI-doctors, AI-robots, AI-hospitals, and AI-cloud-doctors, can
be used effectively in the following fields:
·
medical
interventions (surgery), pharmaceutical production and prescription
(pharmaceutics and pharmacology),
·
immunotherapy and herbal medicine, epidemic
epidemiology, etc.
·
A
cyborg AI doctor
·
AI-robot/AI-hospital/AI-cloud
doctor supporting a human physician, or
·
an
autonomous and remotely controlled AI-robot/AI-hospital are the three various
ways AI can function.
The chances of interacting
with various types of AI appear good. An AI-robot assistance, for instance,
could be useful to a human surgeon performing an intervention using a
remote-controlled machine with surgical instruments. AI can quickly access a
patient's medical history during both surgical and diagnostic procedures and
assess the variables that may influence the choice of treatment (climatic
circumstances, epidemiological situation, the patient's genetic susceptibility
to infections, etc.).
Robotic surgery and
pharmaceuticals, which would enable lowering the expenses of staffing and
24-hour patient care, are the future of medicine. Even a medication driven by
mediocre AI has the potential to help a patient by inducing the placebo effect.
Furthermore, due to the fact that AI's algorithms are created to minimize
errors, patients will be more inclined to believe that the work of AI is
error-free on a subconscious level.
III
CHALLENGES IN THE USE OF ARTIFICIAL INTELLIGENCE IN HEALTH CARE AND MEDICINE.
The application of AI in clinical healthcare practice has enormous
potential to improve it, as the previous section suggested, but it also poses
many ethical and legal issues when used in this sector that are necessary to be
examined.
1.
Consent for use-
The patient-physician interaction will change
as a result of health AI applications in areas like imaging, diagnosis, and
surgery. But how will the application of AI to health care interact with the
concepts of informed consent? Although informed consent will be one of the most
urgent obstacles to successfully Integrating into clinical practice, this
burning question has not earned enough attention in the ethical discussion. It
is necessary to investigate when the informed consent principles should be used
in the context of clinical AI.[9]
How much of a duty does clinicians
have to inform patients on the intricacies of AI, especially the type of ML
used? The data inputs used, the system, and the potential for biases
or are there any other flaws in the data being used. In what
conditions must the patient be informed that artificial intelligence is
being employed at all?
These inquiries are particularly
difficult to respond to when the AI uses "black-box" algorithms,
which may be the consequence of imprecise machine-learning methods that are
highly complex for therapists to completely comprehend. For instance, Corti's
algorithms[10] are
"black boxes" since not even the software's creator is aware of how
the program determines when to notify emergency dispatchers that a person is
having a cardiac arrest. Medical practitioners might be concerned about this
lack of understanding. How much should a clinician, for instance, explain that
they are unable to completely understand the AI's diagnostic or therapy
suggestions? How much openness is required? How does this interact with the
GDPR's purported "right to explanation"? What about situations when
the patient would be hesitant to consent to the use of specific data categories
(such genetic information and family history)? How can we effectively strike a
balance between patient privacy and the security and efficiency of AI?
AI healthcare apps and chatbots are
also being utilized more often in a variety of health-related applications,
such as diet advice, health evaluations, assistance with medication adherence,
and analysis of data gathered by wearable sensors. These applications
raise concerns from bioethicists regarding user agreements and how they
relate to conscious consent as opposed to the conventional. A user agreement is
a contract that a person signs following the informed consent
process without a face-to-face conversation. Most individuals habitually
disregard user contracts because they don't spend the time to study them.[11] People's
ability to adhere to the terms of service they have accepted is further
complicated by the software's frequent updates. What details should be provided
to users of these apps and chatbots? Do users fully comprehend that continuing
to use the AI health app or chatbot may require agreeing to modified terms of
use? How much should informed consent documents mirror user agreements?[12]
2.
Safety and Transparency-
One of the main obstacles for AI in healthcare
is safety. To give an example that has received a lot of attention, IBM Watson
for Oncology employs AI algorithms to analyse data from patient medical records
and assist doctors in looking into cancer therapy options for their patients.
It has recently drawn criticism, meanwhile, for allegedly making "unsafe
and erroneous" suggestions regarding cancer treatments. The issue appears
to be with Watson for Oncology's training process, which only used a small
number of "synthetic" cancer cases developed by Memorial Sloan
Kettering (MSK) Cancer Center physicians as opposed to genuine patient data.[13]
The field has received negative
publicity as a result of this actual instance. It also demonstrates how crucial
it is that AIs be reliable and efficient. But how can we make sure
that AIs maintain their word? Stakeholders, in particular AI developers,
must ensure two critical factors in order to fully utilize AI's potential: (1)
the authenticity and trustworthiness of the datasets; and (2) transparency.
The employed datasets must first be
valid and dependable. When it comes to AI, the saying "garbage in, trash
out" is applicable. The performance of the AI will improve with greater
training data or the labelled data. To get reliable results, the
algorithms frequently need to be improved further. Data sharing is a
significant problem as well. For self-driving cars, for example, where the AI
needs to have a high level of confidence, more data sharing will be required.[14]
Second, some degree of transparency must be
guaranteed in the interest of patient confidence and safety. While in an ideal
world all information and algorithms would be accessible to the public, there
may be some real concerns about safeguarding investments and intellectual
property as well as avoiding an increase in cybersecurity risk. Additionally,
AI developers must be sufficiently transparent, for instance, regarding the
types of data used and any software flaws such as data bias. We should draw
conclusions from cases like Watson for Oncology, in which IBM concealed
Watson's risky and ineffective therapy suggestions for more than a year.[15] Finally,
trust among all parties, especially physicians and patients, is built through
transparency, and this trust is essential for a successful application of AI in
clinical practice.[16]
3.
Fairness and Algorithm
Biasness.
AI has the potential to enhance
healthcare not just in high-income settings, but also by
"globalizing" it and making it accessible to even remote locations.
Any ML system or human-trained algorithm, however, will only be as reliable,
efficient, and equitable as the data it is taught with. AI is also susceptible
to biases, which could lead to prejudice.[17]
Therefore, it is crucial that AI developers are conscious of this danger and
take steps to reduce any potential biases at every phase of the product
development process. When choosing (1) the ML procedures they want to employ to
train the algorithms and (2) what datasets including evaluating their quality
and diversity, they want to utilize for the programming, they should pay
particular attention to the risk for biases.
Algorithms can display biases that
can lead to unfairness with regard to ethnic origins, skin colour, or gender,
as shown by a number of real-world incidents. Biases might also exist with
reference to different characteristics, including age or a disability.[18] Such
prejudices have numerous, varied, and complex justifications. For instance, they
might be the outcome of the datasets themselves (which are not representative),
the selection and analysis of the data by data scientists and ML systems, the
use of AI, etc.
Biased AI could, for example, result
in incorrect diagnoses, render therapies ineffective for specific subgroups,
endanger their safety, and so forth in the medical sector when phenotype- and
occasionally genotype-related information is involved. Consider an AI-based
clinical decision support (CDS) tool that enables doctors to help patients with
skin cancer receive the optimal care. The algorithm, however, was primarily
trained on patients who were Caucasian. As a result, the AI software will
probably provide less accurate or even incorrect suggestions for subpopulations
like African Americans for which the training data was underinclusive.
4.
Data Privacy-
The Royal Free NHS Foundation Trust
was found to have violated the UK Data Protection Act 1998 when it gave Google
DeepMind access to the personal information of about 1.6 million patients,
according to a decision made in July 2017 by the UK Information Commissioner's
Office (ICO). Data exchange took place for "Streams", an app
that intends to assist with the diagnosis and identification of acute kidney
injury, throughout the clinical safety testing phase. Patients weren't properly
informed about how their data was processed during the exam, though. Elizabeth
Denham of the Information Commissioner properly noted that "the erosion of
fundamental privacy rights does not need to be the price of innovation."[19]
Recent case studies demonstrating
patient privacy issues in the context of data sharing and the application of AI
include the legal proceeding Dinerstein v. Google[20] and
Project Nightingale by Google and Ascension[21].
What about who owns the data, though? Health data can be worth billions of
dollars, but some data indicates that the public is uneasy about businesses or
the government making money off of selling patient data[22].
However, there can be other means by which patients can feel valued than actual
ownership. However, there can be other means by which patients can feel valued
than actual ownership. Enabling instance, the Royal Free NHS Foundation Trust
and Google DeepMind reached an agreement for the Trust to use Streams for free for
the next five years in exchange for providing patient data for the app's
testing. Ownership is not always necessary for reciprocity, but anyone wishing
to utilize patient data must demonstrate how doing so will benefit the health
of the very same patients whose data is being used.
IV.
LAW REGULATING AI
IN THE FIELD OF HEALTH CARE AND MEDICINE.
The chances of
interacting with various types of AI appear good. For instance, a human surgeon
doing an operation while using a remote-controlled device, a n AI-robot
assistance may be useful for those using surgical equipment. lowering the
cost of hiring workers and providing 24/7 patient care. Even inadequate or
subpar AI-driven treatment may be advantageous by inducing the placebo
effect in a person. Additionally, patients will unconsciously be more prone to
view the AI's work is error-free since its algorithms are created to
reduce errors.[23]
A. Legal Aspects of AI
in the field of medicine and health care-
AI and the issue of
trust in the field of Medical Science. The development of a single, trustworthy
digital realm has been made possible by advances in science, technology, and
digital technology. Participants in this area are assumed to have confidence in
the information obtained from it, and as a result, their identification and
authentication take place automatically. Regarding the acceptance of electronic
signatures, the topic of the "space of trust" was initially raised.
The European Union passed Directive 1999/93/EC of the European Parliament and
of the Council of December 13, 1999, on a Community Framework for Electronic
Signatures to address this problem. This directive was later replaced by
Regulation (EU) No 910/2014 of the European Parliament and of the Council of
July 23, 2014, on Electronic Identification and Trust Services for Electronic
Transactions in the Internal Market and Repealing Directive 1999/93/EC.[24]
The Eurasian Economic
Union and other international organizations used a similar strategy.
Information could be electronically transmitted remotely thanks to the
Internet. Through the identification and verification of participants in the
information exchange, the existence of the digital space of trust with regard
to electronic signatures ensures that the information received is trustworthy
and reliable.[25]
In general, any digital
information is decoded by translating binary code—a series of ones and
zeros—into text that can be understood by humans. It seems sensible to utilize the
using the e-signature space of trust as a guide for creating a digital space of
trust in the AI, where uniform identification and permission procedures would
apply to medical AI. The creation of a single digital platform for AI trust is
anticipated to be hindered by various procedures implemented in medical schools
and medical customs around the world (the subjective factor). However, all
medical professionals share a common bond regardless of ethnicity, to treat the
patient or prevent them from becoming ill religion, gender, and color, as well
as the hospital's location. As a result, it is likely that international
medical alliances and organizations (such as the World Health Organization, the
International Federation of Red Cross and Red Crescent Societies, Médecins Sans
Frontières, etc.) will play a significant part in the development of the AI
trust space in the digital sphere.
Doctors, patients, the
government, the state, and civil society as a whole will all recognize the
veracity of the information transmitted related to AI, which is the fundamental
tenet guiding the creation of the unified digital space of trust in medical AI.
B. Laws in European Union.
The following sources of legal
regulation apply at the level of the European Union (EU): The application of AI
technology can be identified.
· The EU Parliament passed Resolution
2015/2103(INL) Civil Law in February 2017. Robotics regulations.[26] Robotics
regulations in Europe AI is founded on Isaac Asimov's rules, which state that:
(1) A robot shall not, through act or omission, cause
(2) A robot must accept human
commands if they do not conflict with the first in order to avoid harm to
humans.
(3) A robot is required to ensure its
safety to the degree that doing so does not violate the one of the two laws.
The terms of the aforementioned
regulation primarily apply to robotics, although it might be assumed that, by
analogy, they also apply to AI technology. The resolution proposes giving AI
robots the status of electronic persons, establishing a European registration
system for "smart" robots, and defining culpability for harm caused
by robotics.
Following this, representatives of 25
European nations, including those outside the European Union, signed a
Declaration of cooperation on artificial intelligence in April 2018. According
to its provisions, the signatory nations committed to working on an integrated
European approach to the development of artificial intelligence, pursuing
cogent national policies to increase the competitiveness of the European Union,
and fostering digital innovation.[27]
The Coordinated Plan for Artificial Intelligence
of December 7, 2018, which offers a European strategy for the development of
robots and AI, was created by the European Commission also in 2018. The
participating States' overarching objective in cooperating is to make sure that
Europe emerges as the world's premier region for the development and
implementation of more advanced, ethical and safe AI.[28]
Additionally, ethical concerns
related to the usage of AI are regulated. In order to promote trustworthy AI,
the European Commission approved Ethics Guidelines for Trustworthy AI in 2019
(Ethics Guidelines for Trustworthy AI. High-Level Expert Group on Artificial
Intelligence 2019). Three requirements for trustworthy AI should be met over
the course of the system's complete life cycle: In order to ensure adherence to
ethical principles and values, it should be (1) lawful,
(2) ethical, and
(3) resilient from both a technical
and social standpoint. Even with the best of intentions, AI systems have the
potential to hurt people unintentionally. The
contents of this document state that respecting human autonomy, preventing
harm, being fair, and being explicable are the four basic ethical
considerations for using AI.
In order to digitalize European society and
preserve competitiveness in cutting-edge technologies, such as robotics and AI,
the EU has also adopted the Digital Europe Strategy Programme as a structural
component of the EU financial consolidation for 2021–2027.[29] This program's main goal is
to promote digital transformation by funding the adoption of emerging
technologies in the most crucial fields, each with a separate budget, such as
high performance computing, artificial intelligence, cyber security, and trust,
as well as the implementation and best use of digital literacy.
C. Laws in Russia
Russia
has been creating the initial iterations of national guidelines in the area of
AI in healthcare since 2020. The Technical Committee for Standardization
"Artificial Intelligence" (TC 164), founded on the model of the
Russian Venture Company, will oversee the development. Thus, by 2027, it is
intended to develop about 50 standards pertaining to the use of AI technology
in healthcare in specific areas, such as general requirements and
classification of AI systems in clinical medicine, radiology and functional
diagnostics, remote monitoring systems, histology, medical decision support
systems, image reconstruction in diagnostics and treatment, big data in
healthcare, medical analytics, and forecasting.
A
distinct type of law (rules of conduct) can be recognized in addition to the
conventional legal sources—a medical custom.
René
David and Camille Jauffret-Spinosi, researchers in comparative law, contend
that although legal customs need to be given legal legitimacy, this does not
preclude them from being regarded as an independent, impartial source of law.[30]
Legal traditions are typically accepted by academics as sources of law on par
with legislative acts (for examples, see Panagiotis Zepos' writings). The works
of Raymond Legeais provide a complete description of the contribution of legal
customs to the system of sources of law in various legal traditions. In line
with this theory, it might be claimed that medical legal traditions recognised
by the government or other public organizations (such as the World Health
Organization) can be seen as sources of AI control.
The
topic is not without debate, though. How do we combine medical conventions,
which have been formed over many years of effort by the medical community
(cyborg-AI-doctor, AI-robot, AI-medical organization, and AI-cloud-doctor),
into the legal framework governing AI technologies? Can such legal practices
evolve in the future as a result of the cooperation between doctors and AI
systems? The short history of AI makes it challenging to provide solutions to
these queries. It is anticipated that the state will play a significant part in
these issues since its competent authorities will have to decide which medical
legal conventions (written or unwritten) may regulate AI in healthcare. The
distinction between legal and illegal medical customs, the latter of which is
not approved by the state and does not serve as a source of legislation, is
crucial.
It is
necessary to standardize and make AI accessible to medical legal customs. To
guarantee the caliber and objectivity of AI, this is essential. The rules and
procedures for using AI technology (such as an AI-hospital, AI-robot, and
cyborg-AI-doctor) will be heavily influenced by the legal system in a
particular nation.
We
will then need to develop a worldwide database of medical legal norms adopted
by all nations taking part in the integration effort, which will serve as the
foundation for global AI (i.e., for an AI-cloud-doctor).[31]
V.
CONCLUSION
Cybersecurity,
informed consent, and high standards of data protection high standards of
safety and effectiveness, algorithmic fairness, resilience and cybersecurity,
appropriate transparency and regulatory oversight, all of these crucial
elements must be taken into consideration and handled in order to successfully
develop an AI-driven healthcare system. In this sense, we must do more than
simply update the existing regulatory frameworks to reflect new technological
advancements. However, it is equally crucial to have political and public
debates that centre on the morality of AI-driven healthcare and its effects on
the workforce and society at large. AI has enormous potential to enhance our
healthcare system, but we can only fully realise that potential if we start now
to address the moral and legal issues we face.
[1]G. Michael, A. Steib, C.D. Wiliam, and F.J. Victoria,
“Monitoring expert system performance using continuous user feedback” 3 Journal
of the American Medical Informatics Association 216 (1996).
[2] What is Artificial Intelligence in Medicine? 2021, available
at <https://www.ibm.com/watson
health/learn/artificial-intelligencemedicine> (last visited on April 20th 2021)
[3] New Ebola Treatment Using Artificial Intelligence, available
at <https://www.atomwise.com/2015/03/24/new-ebolatreatment- using-artificial-intelligence/> (last visited on
22nd October 2021).
[4] Phulan Sarma, S. V. Rana, Bikash Medhi, and Manisha
Naithani, “Emerging role of artificial intelligence in therapeutics for
COVID-19: A systematic review” 10 Journal
of Biomolecular Structure and Dynamics 16 (2020).
[5] A. Sami, “Challenges facing
the detection of colonic polyps: What can deep learning do?” 55 Medicina
Journal 473 (2019).
[6] Wenbo Yang, Jehane Michael Le Grange,
Peng Wang, Wei Huang, and Ye Zhewei,
“Smart healthcare: Making medical care more intelligent” 3 Global
Health Journal 65 (2020).
[7] Eric Campo, Daniel Esteve, and Jean-Yves
Fourniols. “Smart homes—Current features and future perspective” 64 Maturitas
97 (2019).
[8] Sophia AI Reaches Key Milestone by Helping to Better
Diagnose 200,000 Patients Worldwide, available at
(last visited on 22 May 2021).
[9] I.G Cohen, R. Amarasingham, A. Shah, B. Xie, “The
legal and ethical concerns that arise from using complex predictive analytics
in health care” 14 Health Aff
1134, available at <https://doi.org/10.1377/hlthaff.2014.0048.> (last visited on 5th 2021).
[10] J. Vincent, “AI that detects cardiac arrests during
emergency calls will be tested across Europe this summer” Verge, available
at <https://www.theverge.com/2018/4/25/17278994/aicardiac- arrest-corti-emergency-call-response 2018> (last
visited March 21st 2021).
[11] I.G, Cohen, A. Pearlman, “Smart pills can transmit
data to your doctors, but what about privacy?” N Scientist, May 2019, available
at <https://www.newscientist.com/article/2180158-smartpills-can-transmit-data-to-your-doctors-but-what-about-privacy>
(last visited on May 16th 2022).
[12] Ibid.
[13] C. Ross, I. Swetlitz, “IBM’s Watson supercomputer
recommended ‘unsafe and incorrect’ cancer treatments, internal documents show”,
STAT, Jan 2017, available at <https://www.statnews. com/2018/07/25/ibm-watson-recommended-unsafe-incorrect-treatments>
(last visited on 7th October 2021).
[14] Figure Eight. What is training data?, available at
<https://www.figure-eight.com/resources/whatis- training-data; 2020> (last visited on Jan 1st
2021).
[15] Ibid.
[16] J. Brown, “IBM Watson reportedly recommended cancer
treatments that were unsafe and incorrect”, Gizmodo, May 2019, available
at <https://gizmodo.com/ibm-watson-reportedly-recommended- cancer-treatments-tha-1827868882; 2018> (last
visited on Feb. 23rd 2021).
[17] B. Wahl, A. Cossy-Gantner, S. Germann, “Artificial
intelligence (AI)and global health: how can AI contribute to health in
resource-poor settings?”, BMJ Glob
Health, March 2018, available at (last visited on July 21st 2021).
[18] N. Sharkey, “The impact of gender and race bias in
AI”, Humanitarian Law Policy, Sept 2018, available at <https://blogs.icrc.org/law-and-policy/2018/08/28/impact-gender-race-bias-ai>
(last visited on Jan. 3rd 2021).
[19] ICO. Royal Free, “Google DeepMind trial failed to
comply with data protection law”, available
at <https://ico.org.uk/about-the-ico/news-and-events/news-and-blogs/2017/07/
royal-free-google-deepmind-trial-failed-to-comply-with-data-protection-law>
(last visited on Oct. 13, 2021).
[21] R. Copeland, “Google’s ‘Project Nightingale’ gathers
personal health data on millions of Americans”, Nightingalegathers, May
2018, available at <https://www.wsj.com/articles/google-s-secret-project-nightingalegathers-
personal-health-data-on-millions-of-americans-1157349679> (last visited on
March 23rd 2021)
[22] S. Gerke, I.G. Cohen, “Potential liability for
physicians using artificial intelligence”
JAMA, May 2019, available at <https://doi.org/ 10.1001/jama.2019.15064.> (last visited on July 21st
2021).
[23] Sophia AI Reaches Key Milestone by Helping to Better
Diagnose 200,000 Patients Worldwide, 2018. Available at <https://www.prnewswire.com/news-releases/sophia-ai-reaches-key-milestone-by-helping-to-better-diagnose-200000-patients-worldwide-680907791.html>
(last visited on March 21st 2021).
[24] Official Journal of the European Union, available
at <https://eur-lex.europa.eu/legal-content/EN/ TXT/PDF/?uri=OJ:L:2000:013:FULL&from=EN> (last
visited on May 15th 2021).
[25] Treaty on the Eurasian Economic Union, 2014, available
at https://www.un.org/en/ga/sixth/70/docs/treaty_on_eeu.pdf (last visited on 20th October 2021).
[26] European Parliament Resolution of 16 February 2017
with Recommendations to the Commission on Civil Law Rules on Robotics, 2015 available
at <http://www.europarl.europa.eu/doceo/document/TA-8-2017-0051_EN.pdf>
(last visited on Jan. 2nd 2021).
[27] EU Declaration of Cooperation on Artificial
Intelligence Signed at Digital Day on 10th April 2018, available at <https://ec.europa.
eu/digital-single-market/en/events/digital-day-2018> (last visited on May 6th
2021).
[28] Coordinated Plan on Artificial Intelligence 2021
Review, 2021, available at <https://digital-strategy.ec.
europa.eu/en/library/coordinated-plan-artificial-intelligence-2021-review>
(last visited on Nov. 23rd 2021).
[29] The Digital Europe Programme for the Period
2021–2027, 2018. Available at <https://eur-lex.europa.eu/legalcontent/ EN/ALL/?uri=CELEX:52018PC0434> (last visited on
Sept. 5th 2020).
[30] David, Rene, and Camille Jauffret Spinosi, Les
Grands Systèmes de Droit Contemporains 107 (The Major Contemporary Legal
Systems, Paris, 10th edn. 2002).
[31] Ibid.
Article Information
CHALLENGES OF ARTIFICIAL INTELLIGENCE IN HEALTH CARE SYSTEM: ITS LEGAL AND ETHICAL CONSEQUENCES.
Authors: HRISHIKESH RAJKHOWA
- Journal IJLRA
- ISSN 2582-6433
- Published 2023/04/29
- Volume 2
- Issue 7
About Journal
International Journal for Legal Research and Analysis
- Abbreviation IJLRA
- ISSN 2582-6433
- Access Open Access
- License CC 4.0
All research articles published in International Journal for Legal Research and Analysis are open access and available to read, download and share, subject to proper citation of the original work.
Disclaimer: The opinions expressed in this publication are those of the authors and do not necessarily reflect the views of International Journal for Legal Research and Analysis.