20221211 DoH Expert Meeting-Halila

14/12/2022
1
Big Data and Augmented
Intelligence: Understanding
Possible Harms for Research
Subjects
RITVA HALILA
M.D., PH.D., DOCENT IN MEDICAL ETHICS
1
Conflicts of interests / interests
u No salaries from the medicinal industry
u No bank accounts abroad
u Work for the Finnish Government and
Public sector à 2022
u Activities: CoE (Chair DH-BIO/ CDBIO
2021-2022, member since 1999
u EU: ERC-Ethics panel 2021-2022,
EGE 2011-2016
u Member of research ethics committees
Medical Doctor, Ph.D., Pediatrician
2
14/12/2022
2
AI and Big Data
u Artificial intelligence:
u the theory and development of
computer systems able to perform
tasks that normally require human
intelligence, such as visual
perception, speech recognition,
decision-making, and translation
between languages (Wikipedia)
u simulation of human intelligence
processes by machines
(Techtarget.com)
u Big Data (Wikipedia)
u refers to data sets that are too
large or complex to be dealt with
by traditional data-processing
application software.
u Data with many fields and from
many sources
u Volume, velocity, variety, veracity
3
Strategic Action Plan on human rights
and technologies in biomedicine
u the governance of technologies and the
strategic objective of “Embedding human
rights in the development of technologies
which have an application in the field of
biomedicine
AI:
u Expert report 2022
u Establishment of a working group 2022
u Report on human rights and AI end 2024
4
14/12/2022
3
Ethical challenges of AI
5
Ethical challenges of Big Data
u data with higher complexity may
lead to a higher false discovery
rate
u capturing data
u Data storage, Data analysis,
search, sharing, updating,
sampling
u Information privacy
u Analysis needs expertise; may
produce risks that exceed an
organization´s capacity to create
and capture value from big data
(Wikipedia)
6
14/12/2022
4
Report on AI in doctor-patient
relationship
u AI systems /doctor-patient relationship
/human rights principles.
u AI according to six themes:
u (1) Inequality in access to high quality
healthcare;
u (2) Transparency to health professionals and
patients;
u (3) Risk of social bias in AI systems;
u (4) Dilution of the patient’s account of well-
being;
u (5) Risk of automation bias, de-skilling, and
displaced liability; and
u (6) Impact on the right to privacy
u https://www.coe.int/en/web/bioethics/deve
loping-a-report-on-the-application-of-ai-in-
healthcare-in-particular-regarding-its-
impact-on-the-doctor-patient-relationship
7
Inequality in health care
u not immediate or universal across
healthcare systems
u Impact on the doctor-patient
relationship in areas suffering from
shortages
u geographical bias and inequalities
in access to high quality care will
be created
u may be more efficient, but also
provide lower quality care
featuring fewer face-to-face
interactions
u depends largely on the role given
to the AI
8
14/12/2022
5
Article 4 of the Oviedo Convention
u Any reduction in oversight or clinical could potentially be viewed as
a violation of Article 4.
u Careful consideration must be given to the role played by
healthcare professions bound by professional standards when
incorporating AI systems that interact directly with patients.
Article 4 – Professional standards:
Any intervention in the health field, including research, must be carried
out in accordance with relevant professional obligations and standards.
9
Transparency to health
professionals and patients
u Easily understandable information
is necesssary for decision making
u Decision making and
interpretations of AI systems are
not transparent and
understandable to those who use
them
u A doctor as a mediator – or not
(i.e, chatbox) both have
challenges
u How to guarantee Interpretability,
traceability, transparency?
10
14/12/2022
6
Risk of social bias in AI systems;
u Technical reasons
u Mismatch between training and testing environments
u Systems developed reflect values and regulations of
manufacturers
u Reflect underlying social biases and inequalities
u Datasets do not represent the targeted population
u Limitations on resources, access or motivation
u Clinical trials as examples: done on white males
u social biases can lead to unequal distribution of
outcomes across populations or protected
demographic groups
u Not easy to detect biases and impairments
u requires careful examination steps to improvements
u Combatting social bias is a multifaceted challenge
11
Dilution of the patient’s account of
well-being
u clinical assessments of patients
based on data (i.e. health apps),
not collected face-to-face?
u How about patient itself having to
say about his/her health and well-
being and his/her view and
context
u Reliance upon data collected by
monitoring technologies as a
primary source of information
about a patient’s health, for
example, can result in ignorance
of aspects of the patient’s health
that cannot easily be monitored. https://toplistbrands.com/top-10-health-apps/
12
14/12/2022
7
Dilution of the patient’s account of
well-being
u Care providers may be less able to
demonstrate understanding,
compassion, and other desirable
traits found within ‘good’ medical
interactions in addition to applying
their knowledge of medicine to
the patient
u AI systems change the
dependencies between clinicians
and patients increasing the
distance between health
professionals and patients thereby
suggesting a loss of opportunities
to develop tacit understanding of
the patient’s health and well-
being
https://www.itnonline.com/article/ai-medicine-way-growth
13
Risk of automation bias, de-skilling,
and displaced liability
u clinicians may trust the outputs or
recommendations of AI systems
not due to proven clinical
efficacy, but rather on the basis of
their perceived objectivity,
accuracy, or complexity
u CoE: “AI-driven health
applications do not replace
human judgement completely
and that thus enabled decisions in
professional health care are
always validated by adequately
trained health professionals.
14
14/12/2022
8
Impact on the right to privacy
u the usage of patient data for training and testing AI systems
u greater development, deployment, and reliance on AI systems in care
may create a greater need to create or curate high-quality real-world
patient datasets to train and test systems
u Innovation can threaten privacy and confidentiality in two ways:
u third party access to (deidentified) patient data and electronic health
records to test and develop AI systems.
u clinicians may be encouraged to prescribe additional tests and analysis not
for their clinical value but rather due to their utility for training or testing AI
systems.
u Risks for rising costs for healthcare
u exposure of patients to unnecessary risks of data leakage or other
breaches of privacy
u Undermines trust between patients and care providers
15
The Oviedo Convention and human
rights principles regarding health
u The Oviedo Convention is designed to “protect the dignity and
identity of all human beings and guarantee everyone, without
discrimination, respect for their integrity and other rights and
fundamental freedoms with regard to the application of biology
and medicine”
u the concept of dignity, identity and integrity of human
beings/individuals should be both the basis and the umbrella for all
other principles and notions that were to be included in the
Convention
u rights to life, physical integrity and privacy, prohibition of
discrimination
u Primacy of the patient over the science and the society
16
14/12/2022
9
u Thank You all
ritva.halila@fimnet.fi
17