Scientists at Novosibirsk State Technical University (NSTU) have created a collection of models to improve the process of diagnosing cancer and choose the most effective treatment methods.
A request for development was received from the Scientific Research Institute of Therapy and Preventive Medicine (NII TPM). According to Irina Yakovina, Associate Professor of the Department of Computer Engineering at NSTU-NETI, Candidate of Technical Sciences, at the first stage of cooperation, under the guidance of Margarita Kruchinina, MD, Head of the Gastroenterology Laboratory at the TPM Research Institute, an information profile of the patient was created based on the systematization of a large array of data collected from various medical institutions (analyses, examinations, questionnaires). In the course of research conducted on small groups of patients, scientists have determined which of the methods of data processing and analysis make it possible to identify the most significant signs — the most alarming "bells" signaling the onset of the disease.
"If cancer is detected at an early stage, the chances of successful treatment increase significantly. Therefore, early diagnosis plays a crucial role in the fight against this disease. The doctors were faced with the task of finding reliable indicators that indicate the start of the pathological process. As specialists in the field of data processing, we helped to find among the many indicators markers that signal the onset of the disease, as well as reference values for the analysis of new diagnostic indicators that can clarify the picture of the disease and help to understand whether additional examination is necessary or everything is normal. It is important not only to detect the disease in a timely manner, but also to correctly determine the stage of cancer, so that, if necessary, to review the patient's management tactics and choose the most effective treatment methods," Irina Yakovina said.
According to her, the use of machine learning methods greatly simplifies the comprehensive analysis of a variety of patient's clinical data. For example, an analysis for fatty acids, which provides a detailed picture of metabolic processes, can be quite detailed, and its interpretation is labor—intensive.
The work lasted for several years, many experiments and studies were conducted, in which students, undergraduates and postgraduates of the Faculty of Automation and Computer Engineering of NSTU-NETI took part. The results were published in the form of reports and articles. In particular, new diagnostic possibilities for colorectal cancer have been identified using an optical cell detection system based on dielectrophoresis, and a system of models has been created for the early diagnosis of colorectal cancer based on electrical, viscoelastic parameters of red blood cells, levels of fatty acids in red blood cell membranes, and blood serum.
As a result of the work, a collection of diagnostic models has been obtained that improve the diagnostic process and enable doctors to make more informed decisions about patient treatment.