EUVIMEDEUVIMED entdecken
Crossref

Predicting tuberculosis incidence using machine learning

N. P. Doktorova, I. A. Vasilyeva, L. E. Parolina, N. Yu. Nikolenko, D. A. Kudlay

Meditsinskiy sovet = Medical Council · 14.07.2026 · (9) · 165-173

DOI: 10.21518/ms2026-112 ↗

Kurz gefasst

Fachartikel in der Zeitschrift Meditsinskiy sovet = Medical Council, online erschienen am 14.07.2026. Verfasst von N. P. Doktorova und 4 weiteren Autor:innen. Die Arbeit nennt 31 Literaturangaben.

AbstractBibliografische AngabenZitierenVerwandte Studien

Abstract (Originalsprache)

Introduction

Forecasting the development of the tuberculosis (TB) epidemic is a complex task, as the epidemic process as a phenomenon is multifactorial.

Aim

To create a dynamic model for long-term forecasting of TB incidence in the Russian Federation using machine learning methods for planning and implementing evidence-based preventive measures, taking into account regional characteristics of the epidemic situation. Materials and methods . The development of the dynamic forecast model included the selection of epidemiological, social, demographic, and economic factors with the greatest influence on the development of the TB epidemic using the Delphi expert assessment method; determination of the weight of each factor; ranking of the constituent entities of the Russian Federation by level of economic development, TB incidence, and mortality; and the creation of a mathematical model using machine learning methods. A Gaussian process was chosen as the machine learning model. The model was trained on data from 2010–2016 and validated on data from 2017–2021. The models were implemented in Python 3 using the open-source scikit-learn library and CatBoost.

Results

The model matrix consisted of indicators that have the greatest impact on the development of the TB epidemic: gross regional product, unemployment rate, life expectancy; TB incidence; the proportion of newly diagnosed pulmonary TB patients with bacterial isolation detected by sputum smear microscopy and culture, the proportion of primary multidrug-resistant TB cases, and the number of TB-related deaths during the first year of observation. The resulting dynamic model of TB incidence using machine learning methods has high predictive ability (R2 = 0.78).

Conclusions

The dynamic model for predicting TB incidence has high predictive capabilities regardless of the level of economic development of a constituent entity of the Russian Federation and can serve as the basis for developing individualized algorithms for selecting measures to optimize the detection, treatment, and prevention of TB.

Zitieren

APA

Doktorova, N. P., Vasilyeva, I. A., Parolina, L. E., Nikolenko, N. Y., & Kudlay, D. A. (2026). Predicting tuberculosis incidence using machine learning. Meditsinskiy sovet = Medical Council, 165-173. https://doi.org/10.21518/ms2026-112

Vancouver

Doktorova NP, Vasilyeva IA, Parolina LE, Nikolenko NY, Kudlay DA. Predicting tuberculosis incidence using machine learning. Meditsinskiy sovet = Medical Council. 2026:165-173. doi:10.21518/ms2026-112

BibTeX
@article{Doktorova2026,
  author = {Doktorova, N. P. and Vasilyeva, I. A. and Parolina, L. E. and Nikolenko, N. Yu. and Kudlay, D. A.},
  title = {Predicting tuberculosis incidence using machine learning},
  journal = {Meditsinskiy sovet = Medical Council},
  year = {2026},
  number = {9},
  pages = {165--173},
  doi = {10.21518/ms2026-112}
}
Kostenlos starten →