Naslov
Dynamic informed consent in medical AI applications

Autori
Luka Poslon ; Čartolovni, Anto

Vrsta, podvrsta i kategorija rada
Sažeci sa skupova, neobjavljeni rad, ostalo

Skup
34. Europska konferencija o filozofiji medicine i zdravstvene skrbi

Mjesto i datum
Varšava, Poljska, 24.08.2022. - 27.05.2023

Vrsta sudjelovanja
Predavanje

Vrsta recenzije
Nije recenziran

Ključne riječi
bez ključnih riječi

Sažetak
This contribution seeks to explain the impact of AI applications on informed consent and to show how AI assistance can affect physicians' considerations of the patient's best interests. Informed consent is a critical moral requirement to foster rational decision making, autonomy preservation, and the respect of individuals. Not only should patients receive relevant information, they should also understand it. The moral basis underlying informed consent can be drawn from four main ethical principles: beneficence, non- maleficence, autonomy, and justice. The concept of informed consent presumes patients are able to autonomously assess the risks and benefits to their health. Integrating medical AI applications into decision- making processes supports physicians across many clinical procedures where turnaround is measured in seconds. Physicians are especially confident in relying on AI systems to make a medical diagnosis, prognosis of disease, or treatment recommendation, and this may entail complex interactions with AI to identify data patterns and generate assessments. Applying AI to clinical decision making or clinical practice presents, however, both benefits and ethical risks and challenges. The impact of technological transformations is likely to affect the physician-patient relationship in terms of respect for patient autonomy, protection of personal data or medical interests, and trust. Epistemic challenges associated with the lack of transparency and explicability in the context of informed consent deserve consideration. AI systems are presently characterized as "black boxes" wherein neither the engineers nor the users of the system can fully account for or comprehend how AI reaches decisions or evaluations. Interpretation-based assessments by AI with little transparency and explicability decrease the reliability of the overall diagnostic process. Keeping in mind that the physician's understanding of how AI systems work and the epistemic mandate for believing AI assessments are correct is critical. Under opaque conditions, the physician cannot justify the judgments needed to fully inform patients about the upcoming treatment. The shortfalls in transparency and explicability when using AI in healthcare, coupled with the need to fully inform patients, are challenges that a Dynamic Informed Consent attempts to overcome. It benefits both patients and physicians with a positive outcome, allowing for greater transparency, for patients' responses to be tracked across different time points, enabling in turn more frequent patient participation in clinical and research activities. Dynamic informed consent can safeguard patient autonomy, personal data, or medical interests. It promotes trust in the healthcare setting, while also being personalized and patient-centred. Moreover, it fosters interactivity, enabling patients to consent to new treatment or change their consent choices in real time as their circumstances change. Dynamic informed consent opens up future opportunities for physicians in health care, improving outcomes and reducing costs in the long run. However, dynamic informed consent warrants further consideration of its fundamental purposes, content design, and form in situations where AI is used for diagnostics or other healthcare interventions. As such, the use of dynamic informed consent can serve to strengthen the physician-patient relationship and may become an essential prerequisite for the ethical use of medical AI.

Izvorni jezik
Engleski

Znanstvena područja
Filozofija

POVEZANOST RADA