Ethical Guidelines AI Healthcare 2023 ICMR
Ethical Guidelines AI Healthcare 2023 ICMR
Ethical Guidelines AI Healthcare 2023 ICMR
Prepared by
DHR-ICMR Artificial Intelligence Cell
© 2023, Indian Council of Medical Research
ISBN: 978-93-5811-343-3
Published by:
Designed & Printed at M/s Aravali Printers & Publishers (P) Ltd., W-30, Okhla Industrial Area,
Phase-II, New Delhi-110020, Phone: 47173300, 26388830-32
These guidelines may be followed by all stakeholder, including innovators,
developers, technologists, researchers, healthcare professional, Ethics Committees
(ECs), Institutions, sponsors, and funding agencies involved in research related to
artificial intelligence in biomedical research and healthcare.
Table of Contents
Foreword.......................................................................................................vii
Preface............................................................................................................ix
Acknowledgements.......................................................................................xi
Preamble.........................................................................................................1
Scope...............................................................................................................3
Applications of AI in Healthcare....................................................................3
4.1 Requisites.......................................................................................44
4.2
Essential information for prospective research participants/
responsibility of researchers..........................................................45
PREFACE
Artificial Intelligence (AI) and Machine Learning (ML) are certainly the
game-changer in the space of healthcare. A wide variety of exciting and
future-looking applications of AI/ML techniques and platforms are available
now. Identification of disease, screening, making diagnosis, medical imaging,
intelligent health operations management, personalized medicine to digital
surveillance for public health, outbreak prediction, and drug discovery are
only some of the emerging uses of AI/ML. The central goal for such systems
should be to make the AI-assisted platforms available for the benefit of
largest section of common people with safety and highest precision
possible. Human beings are the target and that makes the goal complex and
demanding. Data quality, data ownership, usability, trust of the patient-doctor
duo on AI software’s effectiveness, risk acceptance by the doctor and the
patient and ethics are among the various challenges that should be taken up
as medical profession increasingly applies AI based solutions.
Patient privacy, confidentiality and ethics are of paramount concerns for this
emerging field of biomedical research and its application. ICMR is the apex
body for conduct of biomedical research and has been at the forefront to set
Ethical Guidelines for Application of Artificial Intelligence in Biomedical Research and Healthcare ix
the standards on ethics in biomedical and health research. The ICMR ethical
guidelines are highly regarded not only in India but in a number of other nations.
Warm wishes,
x Ethical Guidelines for Application of Artificial Intelligence in Biomedical Research and Healthcare
ACKNOWLEDGEMENTS
This document entitled ‘Ethical Guidelines for Application of Artificial
Intelligence in Biomedical Research and Healthcare’ is the result of continuous
efforts of the member of the Drafting Committee of ethical guidelines of AI in
biomedical research and healthcare. I thank each one of them for their valuable
direction and advice during the formulation of this guidelines. In particular, I wish
to express my profound sense of reverence and gratitude to Prof. (Dr.) Narendra
Kumar Arora, Chairman, for his guidance, supervision and leadership.
I would like to express my gratitude to Dr. Rajiv Bahl, Secretary, Department
of Health Research and Director General, Indian Council of Medical Research,
New Delhi, for his support.
I express my sincere thanks to Prof. (Dr.) Balram Bhargava, Former Secretary,
Department of Health Research and Director General, Indian Council of Medical
Research, New Delhi, for his encouragement and keen interest in publishing this
document.
The publication of this document would not have been possible without the
commitment and contribution of a team of scientists from the Indian Council of
Medical Research and ICMR-National Institute of Medical Statistics, New Delhi.
Last but not the least, I would like to thank the project staff from the ICMR AI
Cell for their valuable support at different stages of the preparation and finalization
of this document.
Ethical Guidelines for Application of Artificial Intelligence in Biomedical Research and Healthcare xi
Preamble
Artificial intelligence(AI) is defined as “a system’s ability to correctly interpret
external data and to use those learnings to achieve specific goal and tasks
through flexible adaption”[1]. AI uses complex computer algorithms to emulate
human cognition albeit with far reaching capabilities of analyzing large
datasets. The field of AI is rapidly expanding and has made significant inroads
in almost all aspects of human life, including healthcare. The incorporation
of AI-based tools and techniques is expected to improve healthcare delivery
by making healthcare accessible and affordable and improving the quality
of care provided. For example, Computed Tomography (CT) scans can be
automatically read by AI as well as radiologists[2]. Tuberculosis screening
can be done by AI using Chest X-Rays with comparable performance as
molecular testing[3], and mammography scans can be used to predict the
onset of breast cancer before visual signs appear[4]. As a result, AI for health
has been recognized as one of the core areas by researchers as well as the
governments.
Despite all the potential benefits, adopting AI for health brings to the fore several
ethical, legal, and social concerns, especially as it pertains to its development
and deployment. The field can be broadly guided by well-established
Ethical Guidelines for Application of Artificial Intelligence in Biomedical Research and Healthcare 1
principles of health research but the development, as well as deployment of
AI-based solutions in healthcare, has to deal with several issues, including
those related to data safety, data sharing, data privacy, etc. For example,
AI-based solutions can empower the masses by permitting easy and early
diagnosis and access to health facilities but unsupervised use of such tools
and techniques is potentially risky. It is therefore mandatory to have ethical and
regulatory framework before AI for Health becomes part of health research
and delivery of healthcare. While the general principles related to biomedical
research and healthcare delivery are applicable to AI for health, the field also
has some unique ethical considerations.
The Indian Council of Medical Research (ICMR) has formulated ethical guidance
documents from time to time for promoting ethical and high quality research
in India. The most recent version of ICMR’s National Ethical Guidelines for
Biomedical and Health Research involving human participants, was released
in 2017[5]. In 2019, Central Drugs Standard Control Organization (CDSCO)
also brought out guidelines for evaluation of new drugs and conduct clinical
trials[6]. India has given these guidelines legal status under the new drugs
and clinical trials rules.
2 Ethical Guidelines for Application of Artificial Intelligence in Biomedical Research and Healthcare
There is no standard and commonly accepted term and therefore the term
AI technology has been used to refer to AI technologies, AI applications, AI
models, AI products, AI-driven solutions, or AI-based solutions in the entire
document.
Scope
These guidelines apply to AI based tools created for all biomedical and health
research and applications involving human participants and/or their biological
data. Considering the far-reaching implications of AI-based technologies in
healthcare, these guidelines are applicable to health professionals, technology
developers, researchers, entrepreneurs, hospitals, research institutions,
organization(s), and laypersons who want to utilize health data for biomedical
research and healthcare delivery using AI technology and techniques.
Both healthcare and AI technologies are rapidly advancing and so shall the
associated ethical dimensions. The document therefore shall remain a living
document and undergo refinement periodically.
Applications of AI in Healthcare
The induction of AI into healthcare has the potential to be the solution for
significant challenges faced in the field of healthcare like diagnosis and
screening, therapeutics, preventive treatments, clinical decision making,
public health surveillance, complex data analysis, and predicting disease
outcomes. This list is likely to grow continuously in future.
Ethical Guidelines for Application of Artificial Intelligence in Biomedical Research and Healthcare 3
As per the National Academies of Sciences, Engineering, and Medicine report,
“Postmortem studies have shown that around 10 percent of patient deaths
can be attributed to diagnostic errors. They also reported that the diagnostic
errors account for 6-17 percent of adverse events as per the review of medical
records[7].
Clinical Care
4 Ethical Guidelines for Application of Artificial Intelligence in Biomedical Research and Healthcare
digital monitoring devices like wearable is one of the areas which have shown
significant development in the recent years. This also provides an alternative
for remote monitoring and identifying early signs of disease by healthcare
workers[9].
Ethical Guidelines for Application of Artificial Intelligence in Biomedical Research and Healthcare 5
during the initial times of the COVID-19 pandemic many countries used AI
based methods for early detection and tracing of contacts to monitor the
spread of the disease[14].
Chatbots are one of potential use of AI in mental health. While mental disorders
continue to carry significant social stigma, and many people struggle to
express their thoughts and feelings directly, mental health chatbots provide
an opportunity for individuals who are inhibited to seek direct professional
psychological and psychiatric help to take their first step towards self-care.
6 Ethical Guidelines for Application of Artificial Intelligence in Biomedical Research and Healthcare
In this approach, AI chatbots on mental health can give initial assistance for
people who are not ready for professional or non-professional care, as well as
augment that support in between interactions with psychologists, psychiatrists,
and peers.
Ethical Guidelines for Application of Artificial Intelligence in Biomedical Research and Healthcare 7
it, providing for quick access and preservation of health and treatment history
for speedier decision-making based on a more comprehensive understanding
of the patient’s unique health profile.
8 Ethical Guidelines for Application of Artificial Intelligence in Biomedical Research and Healthcare
SECTION 1
ETHICAL PRINCIPLES FOR AI IN HEALTHCARE
These general ethical principles address most of the ethical aspects of any
biomedical and health research. Nevertheless, AI for health to a large extent
depends on data obtained from human participants and invokes additional
concerns related to potential biases, data handling, interpretation, autonomy,
risk minimization, professional competence, data sharing, and confidentiality.
It is therefore imperative to have an ethical framework that addresses issues
specific to AI for biomedical research and healthcare.
Responsible AI
Ethical Guidelines for Application of Artificial Intelligence in Biomedical Research and Healthcare 9
and informed approach. However, a lack of consensus on how to employ
AI technology ethically has left many medical decision-makers seeking for
guidance.
Validity
Accessibility and Equity
Trustworthiness Ethical
( Data Security and Risk Minimization
Principles of
Trustworthy) and Safety
Medical AI
Accountability and
Liability Collaboration
10 Ethical Guidelines for Application of Artificial Intelligence in Biomedical Research and Healthcare
The ten ethical principles in Fig 1 addresses issues specific to AI for health.
These principles are patient-centric and are expected to guide all the
stakeholders in the development and deployment of responsible and reliable
AI for health. These principles are as follows –
1.1 Autonomy
The ‘Human in The Loop’ (HITL)[15] model of AI technologies gives room for
humans to oversight the functioning and performance of the system. Clinical
decisions made by the AI technology and the physician may be different
and such disparity causes confusion to the user/patient, whether to trust the
clinician or the AI technology. In such cases, the patient should be provided
with both the options. Before introducing any AI technology in healthcare,
consent process is must for all research projects and evaluation programs
patients should be fully informed about the use of AI technologies benefits
and associated physical, psychological and social risks. Patients must have
complete autonomy to choose or reject AI technologies. There should be
effective and transparent monitoring of human values and moral consideration
at all stages of AI development and deployment.
The patient/ participant has the right to refuse consent. There should not be any
coercion from the government/sponsor/researcher/healthcare professional
and all other stakeholders for using such AI technologies. Over-dependency
Ethical Guidelines for Application of Artificial Intelligence in Biomedical Research and Healthcare 11
on AI system for diagnosis and treatment may negatively affect the patient
clinician relationship and autonomy of the patient. Therefore developers,
institutions, hospitals, health systems and linked stakeholders should develop
policies and guidelines to strengthen the autonomy of the participants.
12 Ethical Guidelines for Application of Artificial Intelligence in Biomedical Research and Healthcare
iv. AI technologies are prone to cyber-attacks and can be exploited
to get access to sensitive and private information, thus threatening
the security and confidentiality of patients and their data. It must be
ensured that the data is completely anonymized and offline/delinked
or from the global technology for its final use.
vi. The Ethical Committee (EC) and other stakeholders must ensure
that there is a favorable benefit-risk assessment. The benefit should
outweigh the risk involved. The risk must be justifiable when the
social and scientific value of AI technology is considered.
viii. AI technologies should be built in line with legal and data protection
requirement of the country and with strict adherence to the basic
principles of ethics.
Ethical Guidelines for Application of Artificial Intelligence in Biomedical Research and Healthcare 13
must ensure timely upgradation of security standards. The security
measures taken to protect the patient data must be publicized so
that it can undergo rigorous scrutiny and gain public trust.
xii. Similar to trials for new drugs and products, if new evidence of
unintended harm associated with using an AI technology comes
to light, then the researcher/ manufacturer and all other relevant
stakeholders must inform the concerned ethics committee and data
safety monitoring board. Post introduction also, like phase 4 trials,
rare and new technology related unintended harm is to be monitored
and documented. The patient/ participant must be informed about
the increased risk of harm. They must have the freedom to withdraw
or continue in the study/treatment.
xv. Special care must be given to ensure the safety and security of
the vulnerable population. The composition and quality of training
14 Ethical Guidelines for Application of Artificial Intelligence in Biomedical Research and Healthcare
datasets may not adequately represent the population in which
the AI is intended to be used. The marginalized communities and
vulnerable groups may be underrepresented in the training data sets.
It may lead to distorted or poor performance of AI technologies. The
EC, sponsors, and all other stakeholders must ensure the adequate
representation of the population in the data. Exposing patients to
unnecessary risk is unethical.
xvi. AI technologies that are used for predicting the possibility of acquiring
a disease may predispose the patient/participant to emotional and
psychological stress. This may potentially lead to the stigmatization
of individuals or communities. The researcher and EC must explore
methods for mitigating such harm. The patients/participants must be
fully informed about the probable outcomes of using AI technology
and the chances of stigmatization and harm.
Ethical Guidelines for Application of Artificial Intelligence in Biomedical Research and Healthcare 15
xx. The potential and the extent to which an AI technology can cause
harm must be assessed before the deployment of AI technology
for widespread use. AI technologies must not cause severe bodily
injury or serious emotional distress.
1.3 Trustworthiness
iii. Reliable and valid, both from technical and social perspectives,
to ensure predictability in the results and outcomes of AI-based
solutions when applied in variety of clinical settings. The results thus
obtained also should be in sync with standard assessment tools.
16 Ethical Guidelines for Application of Artificial Intelligence in Biomedical Research and Healthcare
iv. Explainable, i.e., the results and interpretations provided by AI-based
algorithms should be explainable based on scientific plausibility.
It should be possible to understand the logic behind the results
obtained so that AI technology is valid, reliable and responsible.
The lack of information about the decision-making by AI algorithms
has prompted some to label it as a “black box” which can prove a
deterrent to its wider adoption. A well explainable AI-based solution
is expected to improve the confidence of both the patient and the
health professionals.
Ethical Guidelines for Application of Artificial Intelligence in Biomedical Research and Healthcare 17
purposes in cases where undesirable clinical outcomes may arise
out of the inaccurate interpretation and or recommendation by
AI-technologies. Therefore, for the regulation, acceptance, and
deployment of AI technologies transparency, explainability and
functional understanding is necessary. Limitation in transparency of
the system impairs validation, clinical recommendations, and make
it difficult to identify errors and biases.
viii. All AI technologies must comply with legal norms. Developers must
be able to demonstrate and interpret how the AI technology complies
with data and privacy laws. All software/ privacy policy updates in
an already established AI technology must comply with legal norms.
18 Ethical Guidelines for Application of Artificial Intelligence in Biomedical Research and Healthcare
use of their data is of prime importance to the successful and widespread
deployment of AI. Data privacy must aim to prevent unauthorized access,
modification, and/ or loss of personal data. The application of AI to personal
data must not unreasonably curtail people’s real or perceived liberty
[16]. These practices are crucial in the healthcare sector where medical
information represents sensitive data that, if misused, could harm patients
or subject them to discrimination even if it is unintended. Individual patients’
data should preferably be anonymized unless keeping it in an identifiable
format is essential for clinical or research purposes. All algorithms handling
data related to patients must ensure appropriate anonymization before any
form of data sharing. It is important to know that patient identifiers can be
present as “Metadata” and as “on-image” data and both need to be effectively
anonymized. The issues related to the ownership of the data are complex and
vary based on the national or regional laws and regulations. It also depends on
the degree of data anonymization. Since often data for building AI applications
is gathered from multiple diverse sources (e.g., medical and insurance
records, pharmaceutical data, genetic data, social media, GPS data, etc.), it
can potentially become easier to trace that data to patient and (intentionally
or unintentionally) defeat the goals of privacy. Current Data Protection Act
available in India is the IT Act, 2000. According to section 43A, corporate
bodies possessing, dealing with or handling any sensitive personal data, or
information in a computer resource owned, controlled, or operated by it would
be liable to pay damages as compensation to affected persons if they are
negligent in implementing and maintaining reasonable security practices and
procedures to protect sensitive personal data or information[17]. To ensure
privacy and security of health data, the Indian government is bringing a new
healthcare data protection law - Digital Information Security in Healthcare Act
(DISHA) Bill and Personal Data Protection (PDP); these will have binding on
AI technology ethical guidelines[18]. Some of the salient point on data privacy
are mentions below:
Ethical Guidelines for Application of Artificial Intelligence in Biomedical Research and Healthcare 19
i Users should have control over the data that has been collected from
them for the purpose of developing and designing AI technologies
for healthcare. Users should be provided with the provision to
access, modify, or remove such data from AI technology at any point
in time.
20 Ethical Guidelines for Application of Artificial Intelligence in Biomedical Research and Healthcare
taken previously. The consent must contain the nature of data, to
what extent it is being shared, and possible harm that can occur
from sharing data.
Ethical Guidelines for Application of Artificial Intelligence in Biomedical Research and Healthcare 21
for health should always be appropriately supervised. Open source
software must also follow best ethical practices so as to ensure that
ethical consideration are not compromised at any point of time in the
name of innovation.
iii. The concept ‘Human In The Loop’ (HITL) places human beings in a
supervisory role and is more relevant for healthcare purposes. This will
ensure an individualized decision making by the health professionals
keeping the interest of the patient in the center. Adoption of the HITL
principle throughout the development and deployment of AI for
Health also helps in optimal sharing of accountability by the team
involved in development and deployment of AI-based algorithms.
22 Ethical Guidelines for Application of Artificial Intelligence in Biomedical Research and Healthcare
primarily due to flaws in functionality, then the designer, developer,
or manufacturer may be held responsible. If the harm is caused
due to defective implementation of technology, then the end-user
or organization may be held accountable. There should be a clear
understanding and allocation of responsibility before the deployment
of AI technology.
Ethical Guidelines for Application of Artificial Intelligence in Biomedical Research and Healthcare 23
i. Before deploying AI technologies, the possibilities of biases must be
considered, identified and thoroughly scrutinized. Training data must
not have any sampling bias. Such sampling bias may interfere with
data quality and accuracy. Researchers must ensure data quality.
iii. The existence of bias in the data set can potentially affect the
functioning of AI technology. If there is any allegation of discrimination
or indication of bias in an AI technology, the operation of such a
system must be temporarily discontinued. The manufacturer has the
responsibility to eliminate the bias. Demonstration of a bias-free AI
technology with the optimum function before a competent authority
is mandatory for resuming operations.
24 Ethical Guidelines for Application of Artificial Intelligence in Biomedical Research and Healthcare
research and clinical care. In fact, it is prudent to implement a ‘pre-
deployment testing’ process at every new site were AI is being
deployed so that the AI’s performance can be ensured at a local
level. There should be a robust mechanism to oversee data collection
methods, it should be able to check the fairness and inclusiveness
of data collection and should be able to point out inadequacies and
misrepresentations.
Ethical Guidelines for Application of Artificial Intelligence in Biomedical Research and Healthcare 25
ii. AI technologies might lead to discrimination in ways that may be
hidden or which may not align with the fundamental rights of humans.
26 Ethical Guidelines for Application of Artificial Intelligence in Biomedical Research and Healthcare
number of distinct groups. The government and other relevant
stakeholders have to eliminate the already existing digital divide in
society for universal acceptance and usage of newer technologies.
The introduction of newer AI technology should not cause or worsen
digital divide among populations/groups.
1.8 Collaboration
Ethical Guidelines for Application of Artificial Intelligence in Biomedical Research and Healthcare 27
based solutions is likely to improve the yield from this promising technology.
As a result, stronger collaborations must be achieved with the following
principles:
28 Ethical Guidelines for Application of Artificial Intelligence in Biomedical Research and Healthcare
1.9 Non Discrimination and Fairness Principles
In order to refrain from biases and inaccuracies in the algorithms and ensure
quality, the following principles should be followed:
i. The data set used for the training algorithm must be accurate and
representative of the population in which it is used. The researcher
has the responsibility to ensure data quality.
iii. AI should never be used as a tool for exclusion. Special attention must
be given to under-represented and vulnerable groups like children,
ethnic minorities, persons with disabilities, etc. The AI developers
should promote the active inclusion of women and minority groups.
Ethical Guidelines for Application of Artificial Intelligence in Biomedical Research and Healthcare 29
implementing the technology, the option for reversibility of decision
must be integrated with the AI design.
vii. In case of any unfortunate events arise from the malfunctioning of the
AI technology occurs, then there should be an appropriate redressal
mechanism for the victim. The manufacturer must ensure that there
is a provision for proper grievance redressal.
1.10 Validity
30 Ethical Guidelines for Application of Artificial Intelligence in Biomedical Research and Healthcare
SECTION 2
a. Academic/ Researchers
b. Industry/ Sponsors
e. Ethics Committees
f. Government/ Regulators
Ethical Guidelines for Application of Artificial Intelligence in Biomedical Research and Healthcare 31
2.1 Guiding principles during Development phase
32 Ethical Guidelines for Application of Artificial Intelligence in Biomedical Research and Healthcare
vii. The data collection should be limited to only what is necessary, with
defined time limits of storage of the protected data. Surplus data if
collect should be destroyed or can be stored after obtaining proper
informed consent from the participant /patient.
viii. Existing data sets (e.g., hospital record, EMR, EHR, administrative
sources, research data) can be used with due diligence as already
mentioned and following ethical guidelines and safety, security and
confidentiality measure mentioned before and also referred in the
2017 ICMR ethical guidelines[5].
xi. The source of both the training as well testing data should be properly
documented and reviewed by organizational ECs. In case there is
a difference between the actual purpose for data collection and
the objective of the AI technology being developed using this data,
the same should be documented and reviewed. The population on
which the AI technology is intended to be used should be part of the
testing and validation data sets.
xii. Data should include person from wide range of geographical areas
representing different race, ethnicity, gender, socio economic class
and age to reduce bias in training data.
Ethical Guidelines for Application of Artificial Intelligence in Biomedical Research and Healthcare 33
xiii. Consent waiver can be obtained in case of anonymized data for
retrospective studies.
34 Ethical Guidelines for Application of Artificial Intelligence in Biomedical Research and Healthcare
developers of the AI in comparison to gold standard. The process
must also aim at determining whether or not the proposed AI-based
solution is fit-for-purpose.
Ethical Guidelines for Application of Artificial Intelligence in Biomedical Research and Healthcare 35
important to retain the faith of both the healthcare providers and
recipients.
2.3 Guiding principles for clinical and other health related deployment
Deployment of AI technology is the last and most important step in the journey
of the development of these tools and technologies. It must be dealt with
utmost care since poorly designed healthcare deployment of AI can have
significant negative impact on patients and healthcare providers.
i. Health professionals are the most important contact person for the
recipient of healthcare. Irrespective of whether the data is used for
research or patient management or public health decision making,
concerned health professionals should be aware as to how the data
collected from the participants will be used. They also should be
aware of the processes used for data safety and privacy while it
is being used to train or validate various AI-technology algorithms.
Proper training should be provided to the healthcare providers
regarding the appropriate and safe use of AI technology.
36 Ethical Guidelines for Application of Artificial Intelligence in Biomedical Research and Healthcare
technology is performing optimally or as expected on the target
population.
vii. If any adverse event or injury occurs due to the use of AI technology,
then the user/participant has the right to receive appropriate
compensation. The onus of providing the compensation lies with all
stakeholders.
Ethical Guidelines for Application of Artificial Intelligence in Biomedical Research and Healthcare 37
viii. Health professionals should be aware of potential lacunae and
limitations of the AI technology prior to clinical/research/public
health deployment. Such limitations must be communicated with the
patient/participant.
38 Ethical Guidelines for Application of Artificial Intelligence in Biomedical Research and Healthcare
disclaimer in this regard must be included in the resultant document
highlighting to what extent and for what purpose the said AI tool was
used during biomedical research or clinical decision making. The
terms of service need to include pointers to guide end users that it
is an AI technology and the results/diagnosis/interpretations are not
done by humans. A statement of disclaimer in this regard should
be explicitly conveyed to the health professional and the subject for
whom the technology has been employed.
xv. The terms of service should clearly mention the technology used
behind the solution is an AI technology, validations/certifications if
acquired any and underlying assumptions and disclaimers.
Ethical Guidelines for Application of Artificial Intelligence in Biomedical Research and Healthcare 39
SECTION 3
ETHICAL REVIEW PROCEDURES IN MEDICAL AI
The EC is responsible for assessing both the scientific rigor and ethical aspects
of all health research and should ensure that the proposal is scientifically sound
and weigh all potential risks and benefits for the population where the research
is being carried out. ECs should check the proposals for data source, quality,
safety, anonymization, and/or data piracy, data selection biases, participant
protection, payment of compensation, possibility of stigmatization and others.
40 Ethical Guidelines for Application of Artificial Intelligence in Biomedical Research and Healthcare
of conflicts in ethical requirements during implementation of key ethical
requirements, decisions on the tradeoff should be evaluated regularly.
Ethical Guidelines for Application of Artificial Intelligence in Biomedical Research and Healthcare 41
5. Recruitment of research 5. Selection of Training and Testing
participants (Retrospective/ populations
Prospective study)
42 Ethical Guidelines for Application of Artificial Intelligence in Biomedical Research and Healthcare
14. Community considerations
For further details refer the National Ethical Guidelines for Biomedical and
Health Research involving Biomedical and Health Research, 2017.
Ethical Guidelines for Application of Artificial Intelligence in Biomedical Research and Healthcare 43
SECTION 4
INFORMED CONSENT PROCESS
The researcher must obtain written informed consent from the study participant
for any health research involving human participants and their data. This
requirement is based on the principle that competent individuals are entitled
to choose freely whether or not to participate or continue to participate in the
research and development. Informed consent is a process which primarily
involves three major components providing relevant information to prospective
participants, ensuring the individual’s competence, ensuring the information
is easily understood by the participants and ensuring the voluntariness of
participation.
4.1 Requisites
The document must inform that this is a “research”. The participant must
be able to comprehend the proposed research or system to be developed,
be able to make an informed decision about whether or not to participate
in the study and communicate her/his decision to the researcher in order to
give consent. The consent should be freely given and not obtained through
duress or coercion of any kind, or by offering undue inducements. In case the
participant is not competent (medically or legally) to give consent, the consent
must be taken from a legally authorized representative. Before beginning
any study-related procedures involving the participant, a researcher must
obtain consent. At all stages, participants’ privacy and confidentiality must be
protected. The potential consequences of breech in privacy should also be
mentioned in the informed consent document.
44 Ethical Guidelines for Application of Artificial Intelligence in Biomedical Research and Healthcare
4.2
Essential information for prospective research participants/
Responsibility of researchers
ii. The Researcher must ensure that the patient/research subject has
understood the process and must evaluate the research subject by
“teach-back”, or “show-me” or any other evaluation technique. The
patient/research participant education is a very important step for
an ongoing process which would be required including if consent
obtaining is required multiple times to reduce patient drop-outs. This
should also be documented.
Ethical Guidelines for Application of Artificial Intelligence in Biomedical Research and Healthcare 45
informed consent process, electronic consent and specific issues in
clinical trials may be followed as per the National Ethical Guidelines
for bio-medical and health research, 2017.
vi. IEC approval is required for the use of retrospective data for
developing and validating the AI technology.
46 Ethical Guidelines for Application of Artificial Intelligence in Biomedical Research and Healthcare
SECTION 5
GOVERNANCE OF AI TECHNOLOGY USE FOR
HEALTHCARE AND RESEARCH
Ethical Guidelines for Application of Artificial Intelligence in Biomedical Research and Healthcare 47
State eHealth Authorities and Health Information Exchanges to standardize
and regulate the processes related to collection, storing, transmission and use
of digital health data; and to ensure reliability, data privacy, confidentiality and
security of digital health data”[28].
48 Ethical Guidelines for Application of Artificial Intelligence in Biomedical Research and Healthcare
Ethics Checklist of AI for Biomedical Research and
Healthcare
Table 2: Ethics checklist of AI for biomedical research and healthcare[30]
Ethical Guidelines for Application of Artificial Intelligence in Biomedical Research and Healthcare 49
7. Risks The anticipated physical/social/psychological
involved and discomforts/ risk to participants if any may be
management mentioned and risk management strategy.
strategy
50 Ethical Guidelines for Application of Artificial Intelligence in Biomedical Research and Healthcare
14. Data collection, Describe the source of data, data collection
storage, sharing method and storage, data access procedures,
and access to and data security
data
16. Right to be Can a person retrieve and erase all their records
forgotten or not? And if yes then how? Describe the
process.
Ethical Guidelines for Application of Artificial Intelligence in Biomedical Research and Healthcare 51
References:
[1] M. Haenlein, A. Kaplan, C. W. Tan, and P. Zhang, “Artificial intelligence (AI) and
management analytics,” https://doi.org/10.1080/23270012.2019.1699876, vol. 6,
no. 4, pp. 341–343, Oct. 2019, doi: 10.1080/23270012.2019.1699876.
[2] S. Chilamkurthy et al., “Deep learning algorithms for detection of critical findings
in head CT scans: a retrospective study,” Lancet (London, England), vol. 392, no.
10162, pp. 2388–2396, Dec. 2018, doi: 10.1016/S0140-6736(18)31645-3.
[3] Z. Z. Qin et al., “Using artificial intelligence to read chest radiographs for tuberculosis
detection: A multi-site evaluation of the diagnostic accuracy of three deep learning
systems,” Sci. Rep., vol. 9, no. 1, Dec. 2019, doi: 10.1038/S41598-019-51503-3.
[6] “Ministry of Health and Family Welfare.CDSCO. New Drugs and Clinical Trials Rules
2019 [Internet]. 2019. https://cdsco.gov.in/opencms/export/sites/CDSCO_WEB/
Pdf-documents/NewDrugs_CTRules_2019.pdf”.
52 Ethical Guidelines for Application of Artificial Intelligence in Biomedical Research and Healthcare
[9] A. Bohr and K. Memarzadeh, “The rise of artificial intelligence in healthcare
applications,” Artif. Intell. Healthc., p. 25, 2020, doi: 10.1016/B978-0-12-818438-
7.00002-2.
[10] S. Velupillai et al., “Using clinical Natural Language Processing for health outcomes
research: Overview and actionable suggestions for future advances,” J. Biomed.
Inform., vol. 88, pp. 11–19, Dec. 2018, doi: 10.1016/J.JBI.2018.10.005.
[13] “AI in India: A Policy Agenda — The Centre for Internet and Society.” https://cis-
india.org/internet-governance/blog/ai-in-india-a-policy-agenda (accessed Jul. 05,
2022).
[15] “What Is Human in The Loop (HITL) Machine Learning? – BMC Software | Blogs.”
https://www.bmc.com/blogs/hitl-human-in-the-loop/ (accessed Jul. 05, 2022).
[17] Indian Joint Parliamentary Committee, “The Draft Personal Data Protection Bill.,”
2018.
Ethical Guidelines for Application of Artificial Intelligence in Biomedical Research and Healthcare 53
[18] “Information Technology Act, 2000 - Wikipedia.”
[19] J. Park et al., “A new age-based formula for estimating weight of Korean
children,” Resuscitation, vol. 83, no. 9, pp. 1129–1134, Sep. 2012, doi: 10.1016/J.
RESUSCITATION.2012.01.023.
[20] S. Cruz Rivera et al., “Guidelines for clinical trial protocols for interventions involving
artificial intelligence: the SPIRIT-AI extension,” Nat. Med., vol. 26, no. 9, pp. 1351–
1363, Sep. 2020, doi: 10.1038/S41591-020-1037-7.
[26] “Ministry of Health and Family Welfare, Government of India. National Health Policy
– 2017. New Delhi: MoHFW;,” 2017.
54 Ethical Guidelines for Application of Artificial Intelligence in Biomedical Research and Healthcare
[27] “Final Report on National Digital Health Blueprint (NDHB) | Ministry of Health and
Family Welfare | GOI.”
[28] Government of lndia Ministry of Health & Family welfare (eHealth Section), “DISHA
or the Digital Information Security in Healthcare Act.” Nov. 2017.
Ethical Guidelines for Application of Artificial Intelligence in Biomedical Research and Healthcare 55
Abbreviations:
AI - Artificial Intelligence
DL - Deep Learning
EC - Ethics Committee
56 Ethical Guidelines for Application of Artificial Intelligence in Biomedical Research and Healthcare
MoU - Memorandum of Understanding
Ethical Guidelines for Application of Artificial Intelligence in Biomedical Research and Healthcare 57
Glossary:
1. Algorithms: Algorithm refers to a set of rules/instructions that step-
by-step define how a work is to be executed upon in order to get the
expected results.
58 Ethical Guidelines for Application of Artificial Intelligence in Biomedical Research and Healthcare
7. Human in the Loop (HITL): Human-in-the-loop (HITL) is a branch of
artificial intelligence that leverages both human and machine intelligence
to create machine learning models. In this people are involved in a
virtuous circle where they train, tune, and test a particular algorithm.
Ethical Guidelines for Application of Artificial Intelligence in Biomedical Research and Healthcare 59
Expert Members:
Dr. N.K. Arora, Executive Director, INCLEN Trust, New Delhi (Chairperson)
Dr. M Vishnu Vardhana Rao, Head, AI Cell, ICMR & Director, ICMR-NIMS
New Delhi
Dr. Harpreet Singh, Scientist F and Head BMI Division, ICMR, New Delhi
Dr. Vidur Mahajan, Associate Director and Head of R&D, Mahajan Imaging,
New Delhi
60 Ethical Guidelines for Application of Artificial Intelligence in Biomedical Research and Healthcare
DHR-ICMR AI Cell
Compiled & Edited by:
Mr. Mithlesh Prasad Singh, Scientist B (AI Expert), DHR-ICMR AI Cell, ICMR,
New Delhi
Ms. Harleen Kaur Gulati, Project Research Scientific (Data Expert)-IV, DHR-
ICMR AI Cell, ICMR, New Delhi
Ms. Sonal Khurana, Jr. Data Expert, AI Cell, DHR-ICMR ICMR, New Delhi
Ms. Irin Anna Soloman, Jr. AI Expert, AI Cell, DHR-ICMR ICMR, New Delhi
Ethical Guidelines for Application of Artificial Intelligence in Biomedical Research and Healthcare 61
NOTES
62 Ethical Guidelines for Application of Artificial Intelligence in Biomedical Research and Healthcare
NOTES
Ethical Guidelines for Application of Artificial Intelligence in Biomedical Research and Healthcare 63
NOTES
64 Ethical Guidelines for Application of Artificial Intelligence in Biomedical Research and Healthcare