The paper considers the stage of preparing the database for multi-factor analysis by means of the algebraic model of constructive logic that has been used successfully since 1999 to perform the analysis in medicine and biology. Initial data for model building is a table. Each line in this table is treated as a case where the values of factors and their impacts are marked. The resulting model is represented by a set of result components in the form of factors indicating the limit of detection combined by conjunction sing (pointing to the combined effect). Each resulting component is characterized by the capacity is the essence of the number of lines in the table, which correspond to the specified limits of determining factors in their joint action. The resulting logical expression is characterized by a combination of factors (indicating the detection limits of each of them) in their capacity as the degree of influence on the result. The initial table data should not have contradictions (when the aim is achieved and isn´t achieved by the same values of the factors). To this aim, the program envisages the exception of those targeted lines of which coincide with non-target strings. However, this isn´t always acceptable in cases of a large number of matching target lines and the singular numbers of non-target strings. Then a large number of cases due to the single non-target line are excluded. The analysis of the coincidences of target and non-target lines to select a single non-target line, to remove them from the database on the example of connective tissue dysplasia with magnesium therapy has been proposed. Comparative analysis of the obtained mathematical models was carried out. The effect of improvement of mathematical model on the basis of algebraic model of constructive logic was demonstrated.