Methodological foundation of comprehensive support for dialysis patients based on artificial intelligence technologies

Abstract

The existing software products represented by intelligent decision-support systems based on machine learning algorithms have minimal functionality and solve highly targeted problems in the field of hemodialysis. The aim of the study is to formulate the methodological foundations of the development of a medical information system that allows comprehensively addressing the quality of life problems of dialysis patients through the use of machine learning algorithms. A specific practical goal is to develop an intelligent decision support system for prescribing personalized dialysis and drug therapy for patients with chronic renal failure with an assessment of the long-term risks, as well as to assess the effectiveness of the treatment strategy in terms of the effectiveness of the hemodialysis procedure, the validity of prescriptions for restoring calcium and phosphorus metabolism and antianemic therapy based on the patient's profile. A patient's profile is understood as a combination of socio-demographic characteristics of the patient, functional examinations, laboratory and clinical studies of dynamics monitoring, the "history" of pharmacological prescriptions, and also (partially and not for all patients) genetic studies. The machine learning algorithms include: extreme gradient boosting over decision trees, matching, Cox regression (survival analysis). The significance of this study lies in the universality of the proposed methodology for creating an intelligent decision support system for prescribing personalized therapy to patients. The proposed technique can be used to create a similar medical system for other diseases. An important condition for the methodology translation for evaluating the effectiveness of the treatment of other diseases is the collection of data on patients under the constant supervision of a doctor for a long period of time. This applies to patients with diabetes mellitus, cancer patients, patients suffering from cardiovascular diseases.