A new technique for abdominal surgery prognosis. A neural network approach

The establishment of a correct prognosis in surgical patients constitutes one of the most important issues in medical care. We studied 600 patients from 3 different hospitals applying neural networks in the establishment of prognostic mortality. The "training time", sensitivity, specificity, the percentage of correct classification and area under the ROC curve for each model was evaluated. These systems constitute a useful instrument to manage surgical patients allowing the surgeon to focus is attention in the essential aspects - the patient and the information transmitted by him.