Page 39 - ARNM-3-3

Advances in Radiotherapy
& Nuclear Medicine AI and informed consent in radiation oncology

patients are comfortable with this dual involvement of with disabilities may encounter greater challenges in
humans and AI is crucial for informed consent. While these comprehending AI technologies due to limited digital
technologies hold immense potential, they necessitate a literacy, or language and cultural barriers. To address
re-thinking of traditional consent practices. this, clinicians should adopt tailored, culturally
sensitive communication strategies such as using
2. Ethical challenges in consent taking visual aids, trained interpreters, or simplified language

The use of AI in radiation oncology is transforming to ensure that all patients can meaningfully engage in
clinical decision-making and treatment planning, but it the consent process.
also introduces new ethical challenges, particularly in (vi) Alignment with ethical and legal frameworks:
obtaining informed consent. As AI systems increasingly Traditional frameworks such as the Belmont Report
support tasks such as contouring, treatment optimization, and the Declaration of Helsinki emphasize core
and outcome prediction, patients must now understand principles of respect for autonomy, beneficence, and
not only the physician’s role but also the involvement of informed, voluntary participation - these principles
AI in their care. Several key ethical challenges must be remain foundational even as new technologies emerge.
addressed to ensure consent remains truly informed and In parallel, the World Health Organization’s 2021
13
ethically sound in this evolving context: report on Ethics and Governance of AI for health
(i) Understanding AI’s role: One of the key challenges is underscores the importance of not compromising the
the patient’s understanding of how AI impacts their integrity of informed consent. Legal frameworks are
treatment. Traditional consent processes often involve also evolving alongside ethical considerations. The EU
patients agreeing to a plan based on a physician’s AI Act, a binding regulation, designates AI systems
recommendation. However, with AI now aiding used in healthcare as high-risk, requiring transparency,
decision-making processes, patients must be informed human oversight, and clear communication of system
about the role of AI in their diagnosis and treatment. capabilities and limitations.
(ii) Transparency: Patients may not fully comprehend Illustrative scenario: For example, in adaptive
how AI algorithms make decisions, potentially leading radiotherapy for head and neck cancer, a patient may
to a lack of transparency. It is vital for oncologists to undergo daily imaging where an AI algorithm suggests
explain the AI system’s functions and limitations contour adjustments based on anatomical changes. In this
in simple and understandable terms, ensuring that workflow, the radiation oncologist reviews the AI-generated
patients can make informed decisions. contours before approving the modified plan. During the
(iii) Patient privacy: Protecting patient privacy is a consent process, patients must be informed that AI assists
fundamental concern in the application of AI in in fine-tuning the contours based on real-time data, but the
oncology. The anonymization of data containing clinician retains the responsibility to ensure final approval
sensitive information is imperative to safeguard and treatment safety.
patient confidentiality. This includes demographic
and personal identification details associated with To address these challenges, a revised approach to informed
electronic health records. Furthermore, there is an consent is necessary in the AI-driven era of oncology. This
increasing need to address emerging issues related approach must:
to cybersecurity, data theft, data mining, and the (i) Encourage open communication: Oncologists must
insufficient understanding of third-party access to and engage patients in open conversations about how AI
use of patient data. is utilized in their care, its benefits, and its limitations,
(iv) Bias and accountability: AI systems, while designed to including the possibility of potential errors, as AI relies
be objective, can still be influenced by biases in training on data and algorithms. Explaining the technology’s
data, potentially leading to unequal care. While the capabilities in layman’s terms is essential.
likelihood of AI hallucinations leading to inaccurate (ii) Highlight human oversight: It is crucial to emphasize
or false treatment decisions is low, such errors can still that AI is a tool that complements, but does not
occur and create confusion. It is crucial for treating replace, the expertise and judgment of oncologists.
12
oncologists to discuss these potential limitations and This maintains trust in healthcare professionals and
emphasize the role of human oversight in ensuring reassures patients that they are still at the center
fairness and accuracy in treatment decisions. of decision-making. It should also be emphasized
(v) Equity and accessibility: Receptivity to AI may that AI-generated plans are reviewed, modified,
vary across patient subgroups. Individuals from and approved by radiation oncologists before the
marginalized communities, older adults, or patients treatment.

Volume 3 Issue 3 (2025) 31 doi: 10.36922/ARNM025250030