[Trial of use of computer technology for the assistance of neurologic diagnosis].

This article describes our experience of computer-assisted diagnosis. The original aspects of this work are both the subject chosen and the method used. Our study of clinical and semiological data included the whole of neurological signs above the spinal cord. The interest of the method, based on Bayes theory was in the use of auto-learning and the sample of patients studied as our experience increased. The preliminary results are encouraging, with 85 % correct results, if one considers the first three replies.