Delphic uncertainty and its application in medicine

One of the tasks of statistics is to accurately estimate the uncertainty of inferences (we need to know how unpredictable a given phenomenon is). Such modeling is used to determine insurance premiums, loan repayments, design a health care system, select materials in construction, and publish findings in social sciences or cosmology. Modeling a new type of uncertainty could have applications in medicine.

Traditional medical research is based on randomized controlled trials. In a sterile environment, we administer drugs based on measured patient characteristics, randomly selecting who gets which drug. The research is expensive, so it is done on a small scale—so the information gained is not personalized. This conclusion was supported by the work of Alexander Luedtke, Ekaterina Sadikova, and Ronald C. Kessler (all at the University of Washington).