Risk factors of hospital mortality in patients with TB/HIV co-infection confined in the penitentiary system

Abstract

The problem of tuberculosis with concurrent HIV infection is now becoming increasingly important. It is necessary to determine predictors of a lethal outcome from TB/HIV co-infection for timely selection of the optimal tactics for managing patients and reducing the number of unfavorable outcomes. Objective: to identify significant risk factors of hospital mortality from TB/HIV co-infection among patients in penitentiary institutions. Subjects and methods. For statistical data processing, machine learning methods were used based on the optimal splitting of feature space. To determine the significance coefficients of the identified risk factors, the artificial neural network method was used. Results. The predictive factors of mortality of a tuberculosis patient with concurrent HIV infection undergoing inpatient treatment were identified. The major ones were the following: presence of active HIV associated diseases, CD4 count below 88 cell/mcl, viral load exceeding 595,892.5 copies/mcl, such clinical forms of tuberculosis as bilateral infiltrative tuberculosis with lesion in more than two lobes, disseminated, miliary, fibrous-cavernous tuberculosis, caseous pneumonia, extrapulmonary tuberculosis, and arrhythmia.

Keywords

• туберкулез
• вич-инфекция
• факторы риска
• летальность
• уголовно-исполнительная система
• методы машинного обучения
• цифровая медицина