A system 100 for generating a diagnostic model includes a processor 108 for analysing reference data sets, each including at least one image (Fig. 4； 230, 240). A set may include an end systolic image (minimum left ventricular, LV, volume) and an end diastolic image (maximum LV volume). The analysis identifies at least one feature in each image. The processor calculates a metric depending on the one or more feature. Outcome data, e.g. whether or not the patient develops a cardiac event, associated with at least some of the reference data sets is acquired. A diagnostic model is compiled depending on the calculated metric and the associated outcome data.
