Uncertainty refers to the subjective experience of having insufficient information about a problem of interest.(1;2) A person may be able to reduce uncertainty by collecting informative data. However, often information does not exist in the human domain (question has not been researched) or is not available (not published or not retrievable) – epistemic uncertainty (ignorance).(3) Uncertainty is due to a lack of information – we „don‘t know“! In addition, uncertainty may result from the structure of the environment itself (large number of relevant factors [complexity] or probabilistic nature of the phenomenon of interest) – aleatoric uncertainty (ambiguity).(3) In the latter case, uncertainty is confirmed by the data, i.e. “we are certain to be uncertain”, and uncertainty cannot be reduced by learning. All these sources are relevant for medical diagnosis, prognosis, treatment and prevention.

The concept of risk factors and their assessment refers to a situation with well-defined possible outcomes the probability of which can be measured. In situations of uncertainty, however, such categorization or measurement is not possible.

(1) Han PKJ, Klein WMP, Arora NK. Varieties of uncertainty in health care: a conceptual taxonomy. Med Decis Making 2011;31:828-38.

(2) Fox RC. Medical uncertainty revisited. In: Bendelow G, editor. Gender, Health and Healing. The Public/Private Divide. ed. London: Routledge; 2001.

(3) Hühn J, Huellermeyer E. FR3: A Fuzzy Rule Learner for Inducing Reliable Classifiers. IEEE Transaction on Fuzzy Systems 17(1), 2009.

(4) Huellermeyer E, Brinker K. Learning Valued Preference Structures for Solving Classification Problems. Fuzzy Sets and Systems, 159(18), 2008.