[Reliable diagnostics: clinical gestalt or prediction rule?]

Diagnostic prediction models can support the diagnostic process, both for experienced physicians and for physicians with little experience. More attention should be paid to the incorporation of diagnostic prediction models in the electronic patient record, so that a more accurate probability estimate can be made without simplification to rounded sumscores. A uniform cut-off of sum scores with associated categorization is also undesirable, because it does not take the context of the individual patient sufficiently into account. In the case of a very strong gut feeling, the outcome of a diagnostic prediction model rule alone cannot be sufficient for further policy. Diagnostic prediction models 'only' generate individual objectively estimated probabilities; the clinical decision-making based on these probabilities always needs to be made by the doctor in shared decision making with the patient. Conflict of interest and financial support: none declared.