Diabetes mellitus is still one of the existing pressing global problems that need to be solved. And one of the tools for diagnosing diabetes mellitus can be mathematical modeling methods. This article implements methods for solving inverse problems to restore the parameters of the mathematical model of glucose-C-peptide to determine the volume of natural insulin production and the sensitivity of the pancreas to glucose. The work considers an extended two-chamber mathematical model consisting of 4 ordinary differential equations, 7 parameters and one time-varying parameter, denoting the intravenously administered glucose concentration. The parameters were determined using methods for solving inverse problems using the Levenberg-Marquardt method and additional information about the dynamics of the C-peptide. The case with 10% noise in the data was also considered and the results were compared