AI as an Always-available Oncologist: A Vision for AI-optimized Cancer Therapy Based on Real-time Adaptive Dosing at the Patient Level

This communication presents the long-term vision of AI-optimized cancer therapy based on automated adaptive dosing. The idea is to have an AI-controlled therapeutic system that administers microdoses from information obtained from low-power sensors, which could improve patient quality and survival. While this idea has not been implemented for cancer yet, there are similar health interventions in cancer (not using AI) and in diabetes (using AI) that serve as precedents. However, there are still major challenges to tackle, such as identifying relevant, measurable, and reasonably costly tumor markers and dealing with the enormous combinatorial potential for a rapid and effective response in individual cases. The paper proposes a dual process to address these challenges, involving collecting initial findings in vitro and investigating tumor markers for their transferability to in vivo systems. If successful, this intervention strategy could have vast implications for the treatment of cancer, the second leading cause of death in the world. It is important to consider ethical and regulatory considerations in the development of this strategy.