Meaningful incorporation of artificial intelligence for personalized patient management during cancer: Quantitative imaging, risk assessment, and therapeutic outcomes

Abstract Personalized medicine stands to become the future of health care in many domains, including oncology, which is both a reflection and cause of increasing genetic discoveries and potential decision support models. In this chapter, we focus on research surrounding decision support in cancer, with a lens on the medical journey of patients with cancer. These three areas of focus follow a patient from diagnosis to assigning the best therapies: (1) quantitative imaging, which uses numerical values extracted from images to predict outcomes such as a diagnosis, (2) risk assessment in cancer patients, which assigns a prognosis, and (3) therapeutic outcome prediction, which attempts to assign each patient to their predicted best therapy. In the final section of this chapter, we elaborate on what we believe to be meaningful incorporation of artificial intelligence for personalized cancer care. We urge model creators to be transparent with their workflows, label acquisition, validation, and performance reporting in order to meet the meaningful criteria and therefore increase public confidence in their software for clinical use.