Chronic lymphocytic leukemia prediction using data mining methods

Abstract

Relevance. Chronic lymphocytic leukemia (CLL) is one of the most common lymphoproliferative diseases of the European population with an increase in the elderly and senile age frequency. In this category of patients standard approaches to predicting overall survival do not take into account the presence of comorbid pathology and have low accuracy. In view of this, the search for parameters that affect the overall survival rate of patients with CLL is of particular relevance.The aim of the study is to identify factors affecting the CLL patients overall survival at the stage of CLL diagnosis.Materials and methods. The data of 132 CLL patients with stage A-C according to Binet with known overall survival were retrospectively analyzed. The problem was solved by data mining methods, namely using logical classification algorithms.Results. The glomerular filtration rate is defined as a parameter that objectively justifies the real terms deviation of the patients overall survival from the calculated ones according to the standard Binet staging system. For this parameter, an if…then rule is formed, which makes it possible to predict the patient’s survival. If the GFR value at the time of diagnosis of CLL is more than 76 ml/min /1.73 m2, we can say that the patient will overcome the calculated median survival data for the corresponding stage of CLL according to Binet. Otherwise, the overall survival of the CLL patient will be less than the estimated median survival according to Binet.Conclusion. The analysis of the study allows us to conclude that it is advisable to use data mining methods in predicting the patients overall survival with CLL. The clinical examples given in the article show their effectiveness. According to the study results, an application for invention No. 2022104419 was issued.