Analyse décisionnelle en radiologie

Decision analysis is a technique which allows to clarify in an explicit, probabilistic and quantitative way the possible answers to a problem and to help the decision making process. The creation of algorithms, graphically displayed as decision trees in most cases, requires the introduction of quantitative information of two types: probabilities of the events that result from answering to the initial question, and utilities of the possible outcomes of these events. The choice of the optimal solution is based on the calculation of combinations of these data. Although the construction of models can be complex and time-consuming, their practical use has been simplified by the ease and user-friendliness of available softwares. To date, the applications of decision analysis in medicine have focused on cost-effectiveness studies and on the simulation of randomized trials. At the individual level, clinical applications of decision analysis could provide in a near future an interesting tool for the clinician involved in patient management. Even when doubts remain regarding the reliability of the data entered into the model, further sensitivity analysis provide a convenient way to test the validity of the drawn conclusions. This article hopefully offers a simplified and practical approach to basic decision analysis.