Outliers are those special points that differ significantly from most sample data. It can skew the data and present less accurate prediction results and detecting them is very important for obtaining more accurate predictions.The manuscript aimed to compare several regression models, including the multiple regression model using ordinary least squares (OLS), the quantile regression model (QR), and the ridge regression model (RR), to identify the model with high efficiency in the presence of extreme values in the data, both before and after treating the extreme values in the data. The comparison was applied to regression model of a sample of 62 sugary patients on kafr el- shikh university hospital, to study the effect of blood sugar level(x1), high blood pressure(x2), low blood pressure(x3), and weight (x4) on the cumulative glucose rate(y).  The study was conductedas follows:
1. Outliers were detected in the variables, namely the cumulative glucose rate variable, blood sugar level, low blood pressure, and weight, while no outliers were detected in the high blood pressure variable.This was done by relying on the Box plot.
2. The extreme values in the data were treated using the trimmed mean method.
3. The regression model was estimated in the presence of extreme values and after treating them. It was found that the best regression models before and after treating the data were the quantile regression model, which has the lowest mean squared errors before and after treating the data.