Creation of a differential diagnosis system in the class of non-Hodgkin's lymphomas through the construction and implementation of pattern recognition algorithms

Abstract

The paper considers the most relevant application of the mathematical theory of pattern recognition, its models and methods - differential diagnosis in a selected class of malignant forms of diseases in hematology, such as non-Hodgkin's lymphomas. The construction and software implementation of algorithms for the diagnosis of non-Hodgkin's lymphomas based on clinical signs and symptoms, as well as the subsequent development of decision support software, are proposed. A specialized data set was also generated during the study.; a software prototype was implemented with the possibility of entering clinical signs and obtaining diagnostic output; the developed system was tested on selected data; a comparative analysis of the results obtained using various algorithms was performed. The built diagnostic system has a number of significant advantages and is based on the use of diagnostic features available in routine clinical practice, such as clinical, laboratory and anamnestic data, which eliminates the need for expensive imaging or genomic studies. This system can be implemented in conditions of limited computing and information resources, including regional clinics, as well as adapted to the specifics of local medical institutions by learning from its own data.

Speaker

Matvey Vladimirovich Sedykh
Gagarin's State Technical University of Saratov
Russia

Discussion

Ask question