Common Misconceptions and Future Directions for AI in Medicine: A Physician-Data Scientist Perspective

摘要

'Healthcare is an information industry that continues to think that it is a biological industry.' (Laurence McMahon at the AAHC Thought Leadership Institute meeting, August, 2016) The following are common misconceptions (by both clinicians as well as data scientists) about AI in medicine that are understandable and human: Clinicians will be replaced by AI. There is a fundamental lack of understanding of what clinicians do even amongst august data scientists and seasoned venture capitalists, and this deficiency renders it easy to think that computer vision and interpretation of medical images alone is sufficient to replace image-intensive subspecialties like radiologists, pathologists, ophthalmologists, dermatologists, and cardiologists. The doctor's tasks can be divided into perception (visual image interpretation and integrative data analytics), cognition (creative problem solving and complex decision making), and operation (procedures) (see Fig. 1). The computer is much stronger in perception tasks but not yet facile with cognition nor operation parts of the clinician's tasks. It is also notable that all those who proclaim an end to radiologists are not radiologists (or even physicians) themselves.