INVESTIGATION OF THE EFFECTIVENESS OF VARIOUS MACHINE LEARNING MODELS IN PREDICTING THE PROPERTIES OF EXOGENOUS FLUOROPHORES

10.25712/ASTU.1811-1416.2024.03.006

Abstract

The article analyzes the effectiveness of using various machine learning models to predict the spectral properties of exogenous fluorophores, which are key in the diagnosis of cancer. The application of AI algorithms for the rapid and cost-effective search for new fluorophores contributing to the early detection of cancer is being investigated. The article evaluates the effectiveness of various machine learning models in predicting the properties of exogenous fluorophores used in the diagnosis of cancer. The paper explores the use of artificial intelligence algorithms for the rapid search for new fluorophores that contribute to the early detection of cancer. Special attention is paid to optical biopsy as a non-invasive method of tissue examination for early diagnosis of pathologies. The article summarizes data from the PubChem and GeoMcNamara databases and analyzes the molecular properties of fluorophores and their spectral characteristics. Using machine learning models such as linear regression, the support vector machine method, random forest and XGBoost, the results of radiation wavelength prediction for fluorophore samples were obtained. The results of training and testing of models indicate the high accuracy of the work of XGBoost and Random Forest. The study highlights the importance of developing effective fluorophores for early cancer diagnosis and presents machine learning models as tools for processing and analyzing data in this area, which allows us to focus on the prospects and applicability of advanced research methods in oncology and medical chemistry.