[Towards an automatic classification of psychotropic drugs, based on computerized card-index of their properties]

The authors give the used statistical approach to realize automatic classification on computarized clinical data about 111 psychotropic drugs. Four ways of analyses are used: a reciprocal averaging method to describe and to put together the numerous properties, a cluster analysis and a sequential dynamic clusters method to define groups, a decisionnal method discriminal analysis, to put non classified drugs down these groups. This methodology points out six groups characterized by clinical properties. Others reciprocal averaging analysis part: desinhibitors antipsychotic drugs and sedative antipsychotic drugs, monoamine oxydase inhibitors and others antidepressants, amphetamine drugs and non amphetaminic psychostimulants. This statistical way permits classification of new drugs without a priori.