The development of artificial intelligence in personalized oncology medicine is a significant advance. However, it requires a framework, especially from an ethical perspective. For this reason, we outline in this video the ethical issues surrounding AI development, which will eventually relieve physicians of certain time-consuming tasks, allowing them to dedicate more time to patients.
Feel free to also watch our videos on dynamic consent and on artificial intelligence and personalized medicine in the fight against cancer.
https://youtu.be/E5sITRq1cto
Retranscription de la vidéo
Artificial intelligence has the potential to advance rapidly personalized medicine in oncology.
The effectiveness of AI depends on the amount of data that patients allow access to, as well as the expertise provided by the medical community.
Therefore, the success of this approach relies on understanding AI and gaining acceptance from all stakeholders, while considering the ethical implications of its widespread use.
First and foremost, it is essential to ensure patients that appropriate measures will be taken to safeguard all their rights, particularly regarding privacy and access to information.
The concept of dynamic consent appears to be the best way to empower patients to actively participate in research, while also expanding research possibilities.
For example, the acquired data can be used to train an artificial neural network.
In simple cases, we have the tools to interpret the results.
In the most complex systems as in deep learning, it may become impossible for humans to understand the pathway taken by AI.
The ability to understand algorithmic decisions is called explainability. But who is responsible for decisions without explainability? Are healthcare professionals or health authorities responsible for validating medical products, or the developers and publishers of applications?
For AI to be developed and to contribute to the anticipated medical progress in oncology, the medical community, civil society, and public authorities must discuss all these aspects. Above all, this requires good information. This is why CARPEM teams participated in creating a university diploma in AI applied to health. AI should eventually become highly effective in assisting with diagnosis, prognosis evaluations, and treatment identification.
However, the decision to convey information, the modes and timing of communication and the prescription of the right treatment at the right moment for each patient involve deliberate choices which require consciousness—something that AI completely lacks.
These are choices that belong exclusively to the physician.
Our commitment is to build AI in oncology around and in service of humanity. This AI will relieve the physician of time-consuming tasks that can be automated.
It will allow the physician more time to converse with patients and take the best decisions regarding their health together, respecting their values.