Authors
1
Associate Professor & HOL, Laboratory of Collection, Analysis and Control of Biological Signals, Tomsk State University of Control Systems and Radioelectronics (TUSUR), 634050, Tomsk, Russian Federation
2
Assistant Professor, Security Department, Tomsk State University of Control Systems and Radioelectronics (TUSUR), 634050, Tomsk, Russian Federation
3
Associate Professor, Laboratory of Collection, Analysis and Control of Biological Signals, Tomsk State University of Control Systems and Radioelectronics (TUSUR), 634050, Tomsk, Russian Federation Siberian State Medical University, 634055 Tomsk, Russian Federation
4
Associate Professor, Laboratory of Collection, Analysis and Control of Biological Signals, Tomsk State University of Control Systems and Radioelectronics (TUSUR), 634050, Tomsk, Russian Federation
,
Document Type : Research Article
Abstract
The paper considers the possibility of predicting tumor parameters (weight, volume) depending on the method of treatment and doses of drugs used. A comparison is made of the applicability of several regression models, taking into account the limited amount of data. Brown's adaptive model, paired regression, multiple regression are compared. Previously, all constructed regression models were tested for adequacy by assessing the significance of the coefficient of determination according to Fisher's criterion, and only the models that were recognized as adequate were used for further predictions. The advantage of Brown's model on the considered limited data set is shown experimentally and recommendations are given on the choice of the method parameters.