電気通信システムと管理

Medical Artificial Intelligence

Abstract

Dr. Yifan Zhu

Medical  artificial  intelligence  (AI)  is  a  hot  point  in  AI  utilities  now,  of which clinical diagnosis is one important AI administration. Diagnostic  AI  (D-AI)  is  the  simulation  of  doctor’s  clinical  thinking  for clinical  analysis  and  decision.  These  abilities  come  from  many-years medical education and clinical experience. Therefore, D-AI should also be based   on   medical   knowledge   from   medical   university   and   clinical experience. However,  at  present  most  D-AI  is  based  on  big  data  analysis,  it  is  now right.  The  first  big  question  of  big  data  based  D-AI  is  that  it  is  not evidence-based analysis which is absolutely required by evidence-based medicine in clinical analysis and decision. The second is that its algorithm can   not   be   interpreted   by   medical   knowledge,   thus   could   induce algorithm black box, which is now commonly suggested to avoid. A medical knowledge based D-AI algorithm is needed to set up right D-AI, simulating clinician’s clinical thinking.