PharmaDM - Features and benefits

Features and benefits of PharmaDM technology

data are analysed where they are in a relational database; no re-formatting is required
all data in the database are accessible to the data mining engines; no data reduction step is necessary with consequent loss of information
any property that is stored in the database can be used to explain observations; potentially relevant properties need not (but still can) be chosen up-front
the data mining engine does not distinguish between chemical, biological or clinical properties when searching for relevant ones; this renders truly integrated cross-domain data analysis possible for the first time
the data mining engine exhaustively searches the database for relevant properties and their combinations, thus assuring that an optimum hypothesis accounting for the observations is found
the user can at any time guide or restrict the data mining engine by suggesting or imposing properties to be used in the explanation
the hypotheses generated by the data mining are phrased in a form that is comprehensible to a domain expert, e.g.

    IF the molecule contains a benzene ring 
    with two substituents in meta position, 
    one of which is a methyl group
    THEN the molecule is active
    OR ELSE the molecule is inactive.

background knowledge can easily be incorporated in the analysis; it suffices to add a table to the relational database (e.g. with public domain information about the chemical properties of atoms, or biological information about genes)
by making it easier to interact with complex databases and by generating superior as well as comprehensible hypotheses, much more value can be extracted from databases that you have invested in building