Anticipating the Future of Artificial Intelligence in Medicine and Health Care: A Clinical Data Science Perspective

Abstract

The future of artificial intelligence in health and medicine is promising as a resource to reach both precision medicine and population health, but will depend on the degree of synergy between humans and technology and the extent of convolution between clinicians and data scientists. Artificial intelligence and deep learning will continue to advance at an exponential pace with adoption of imaging technologies such as few shot learning, generative adversarial networks, transfer learning, and other technologies. Much new knowledge in medicine will come this coming decade from unsupervised learning or even self-supervised learning to mitigate the burden of clinicians involved in projects. Deep learning will need to be in the context of a cognitive architecture for future dividends in biomedicine. In addition, artificial intelligence will converge with emerging technologies such as extended reality to engender a digital twin dimension to patient care and medical education and training. Artificial intelligence and its adoption, however, can be slowed by lack of adequate access to biomedical and healthcare data, specifically the lack of disease population data in the form of images, biomarkers, or other phenotypic expressions of disease. Artificial intelligence will also face daunting challenges of clinician adoption as well as legal, regulatory, ethical, and financial challenges. A cohort of dually-trained and educated clinician data scientists can accelerate the aforementioned dimensions as productive liaisons to both domains. The best dividend in the future of artificial intelligence in medicine is, perhaps, a reclaiming of humanity in medicine.