The invention relates to a method and apparatus for predicting a sudden heart abnormality for an individual patient. In order to provide a prediction mechanism that is suitable for acute care, three sub-indices are determined based on medical data obtained from the patient. The first sub-index indicates the level of deterministic chaos in the heart rate variability of the patient, the second sub-index indicates the energy level in the myocardium of the patient, and the third sub-index indicates the degree of ventricular arrhythmia of the patient. Based on the first, second, and third sub-indices, at least one overall risk index is then determined, the overall risk index indicating the risk level of a sudden heart abnormality for the patient.