Abstract
<jats:p>Abstract. Currently, in medical practice, methods and tools are being developed to diagnose diseases and predict the likelihood of developing possible complications. The article presents a method for predicting the development of otogenic intracranial complications in children. Objective. To evaluate the effectiveness of the developed methods for predicting otogenic intracranial complications (OICC) in children using IT technologies. Patients and methods. A clinical study was conducted in which 130 children (80 boys and 50 girls) with acute purulent otitis media, mastoiditis, and otogenic intracranial complications took part. The average age of the patients was 4.32±0.31 years. All patients are divided into 3 groups. Group I included 50 children with acute otitis media (AOM) (30 boys and 20 girls). Group II included 50 children (25 boys and 25 girls) with mastoiditis. Group III included 30 children (15 boys and 15 girls) with otogenic intracranial complications (OICC). The leukocyte shift index (LSI) was chosen as a marker. It was shown that in the age group from 2 to 16 years the development of AOM and mastoiditis is predicted at LSI ≥ 2.08±0.23, and the development of OICC, at LSI ≥ 3.95.08±0.23. To create a predictive model of OVChO, we used the method of machine learning and constructing a decision tree based on LSI. The accuracy of the model showed that the sum of all correctly identified diagnoses is 71%, and each diagnosis has an error probability of no more than 5%. Thus, the developed model allows us to predict the probability of developing OICC in children with diagnostic accuracy of up to 95%. Machine learning can be used to build a decision tree when predicting OICC in children for the purpose of timely surgical treatment and systemic antibacterial therapy.</jats:p>