A Combination of the K-nearest Neighbor Algorithm and the Principal Component Analysis for Classification and Screening of Patients with Chronic Lymphocytic Leukemia and Multiple Myeloma.
Robert V. Butyaev, Daniil A. Chernyshev, Ludmila V. Plotnikova, Alexander M. Polyanichko.
Saint Petersburg State University, Saint Petersburg
Abstract
This study focuses on the development of screening approaches using infrared (IR) spectroscopy for the timely diagnosis of chronic lymphocytic leukemia (CLL), one of the most common oncohematological diseases. Additionally, the effectiveness of the combination of these methods on multiple myeloma (MM) samples will be evaluated. The researchers used IR spectroscopy in combination with machine learning methods to classify blood serum samples with diagnosed CLL and MM. The study involved taking spectra of blood serum samples from CLL, MM patients and healthy donors, and applying the principal component analysis (PCA) and k–nearest neighbor’s algorithm (kNN) for classification. The algorithm correctly classified all analyzed samples in the test set, suggesting its potential for use in a CLL screening system. Additionally, cross-validation was carried out with the division of the original set into four equal parts and the macro recall averaged over four repetitions was evaluated. The macro recall value during cross-validation was 0.85 – this means that there is an 85% probability that the new sample will be classified correctly. The accuracy of the classification of MM patients was about 90-100%. Further research with larger sample sizes is recommended to improve the accuracy of classification and increase the lifespan of CLL and MM patients.
Blood serum samples were kindly provided by the Federal State Budgetary Institution "Russian Research Institute of Hematology and Transfusiology of the Federal Medical and Biological Agency" (St. Petersburg). Part of the work was carried out using the equipment of the Science Park of St. Petersburg State University (Optical and laser methods for the study of matter, Center for Diagnostics of Functional Materials for Medicine, Pharmacology and Nanoelectronics, Cryogenic Department)
File with abstract
Speaker
Robert Butyaev
Saint Petersburg State University, Saint Petersburg
Russia
Discussion
Ask question
