In this paper standardized vectorial (quantitative) representation of medical data is suggested for use in patient records. Vectorial representations are (as sequences of numbers) language independent, precise, directly comparable, and they allow advanced evaluation, e.g. similarity calculation using well defined distance functions. It is possible to search for a patient with a certain combination of diagnostic parameters on the Web records of patients with similar parameters. The information about chosen treatments and treatment outcome at these patients can be used anonymously or pseudonymously for decision support. Because patient records from all countries can be compared, in the long run this could open systematic access to a very large wealth of clinically relevant information. Here the technical principle is described and illustrated by examples (e.g. similarity search of heart sounds). Previously published material is integrated in parts for explanation of the motivation and background.