[New methods for computerized classification of psychopharmaceutical agents (author's transl)].

We describe a statistical technique for achieving unbiased classification of 111 psychopharmaceutical agents from computerized clinical data. Three different methods were used: the reciprocal averaging method describes and pictures the clusterings of the numerous clinical datum available; the cluster analysis method corroborates the arrangement of the drugs and their effects into groups; the discriminating analysis method assigns the remaining drugs to their groups. The drugs were ultimately classified into six different groups. One of these groups, which included many of the antidepressant drugs, was reanalyzed with the reciprocal averaging method. This led to separating the monoamine oxidase inhibitors from the other antidepressants. Our technique will be immediately useful for classifying new psychopharmaceutical agents. If biologic data are included, this technique will be helpful for studying correlations between the structure and the effects of drugs.