1.Multinomial logistic regression is used to model categorical outcome variables. The log odds of the outcomes are modeled as a linear combination of the given predictor variables. It is an extension of normal logistic regression that allows us to have a response variable with more than just two categories.

2.Interpretation of model coefficients: Multinomial model coefficients are given initially as log-odds ratios. These model coefficients can be converted to the ratio of the probability of choosing one outcome category over the reference category by taking the exponentiated value of the original coefficient. This exponentiated value is often referred to as relative risk.

3.Generalized Linear Modeling (GLM) uses the response variable as the continuous tablets outcomes and determines whether allele categorization had a significant effect on tablet consumption.

4.Prior to modeling, the data is checking for outliers, defined as more than 3 standard deviations away from the other sample principal component values along a given axis, using Principal Component Analysis (PCA). 

•For CYP2D6, The coefficient TABLETS was found to be a statistically significant (P < 0.05) contributing factor in the classification of the group 'Poor' when 'Normal' is used as the reference group. The relative risk ratio for a one unit increase of the variable TABLETS is exp(0.4696) = 1.599.

•The COMT classification difference of "High" to "Low" was found to have a significant effect on the response variable "Tablets.”

•The CYP3A4, CYP3A5, and OPRMI classifications did not have a significant effect on the response variable “Tablets”