Neural networks trained with simulation data for outcome prediction in pallidotomy for Parkinson's disease

摘要

Neural networks were trained using the ALOPEX technique to predict the outcome of pallidotomy, an operation performed to alleviate Parkinson's disease. Because of the limited number of operations performed, the training procedure relied heavily on simulation data. The assumptions from which the simulation data were derived were reflected in the performance of the resultant networks; the networks nonetheless retained the flexibility to respond appropriately to unusual inputs.