A systemic approach to the study of a new multi-parameter model of the COVID-19 pandemic spread is proposed, which has the ultimate goal of optimizing the manage parameters of the model. The approach consists of two main parts: 1) an adaptive-compartmental model of the epidemic spread, which is a generalization of the classical SEIR model, and 2) a module for adjusting the parameters of this model from the epidemic data using intelligent optimization methods. Data for testing the proposed approach using the pandemic spread in some regions of the Russian Federation were collected on a daily basis from open sources during the first 130 days of the epidemic, starting in March 2020. For this, a so-called data farm was developed and implemented on a local server (an automated system for collecting, storing and preprocessing data from heterogeneous sources, which, in combination with optimization methods, allows most accurately tune the parameters of the model, thus turning it into an intelligent system to support management decisions). Among all model parameters used, the most important are the rate of infection transmission, the government actions and the population reaction.