Applying machine learning and feature analysis to predict demand in the Russian pharmaceutical market

This article explores the use of computer-based methods for analyzing tabular data to forecast consumption in the Russian pharmaceutical market. It describes the key stage of developing an information system designed to forecast drug procurement and support management decision-making in the pharmaceutical supply chain. It examines the specifics of medical organizations' procurement activities and the key risks associated with planning drug demand and pricing. It details the modern methods used in the study, including machine learning models and feature significance analysis using SHAP. It describes the data preparation and preprocessing process, including collecting, cleaning, transforming, and encoding features, as well as generating training and test samples for building regression models. Particular attention is paid to identifying factors influencing drug pricing and improving forecasting accuracy through the use of specialized models for specific drug groups. The economic impact of implementing the developed tool is assessed. It enables medical organizations to more effectively manage procurement, optimize budgets, reduce financial risks. Specific attention is given to forecasting drug prices and automating the planning and procurement process as part of the sustainable and rational development of the Russian pharmaceutical market.

1. Gusev A.V., Zarubina T.V. Clinical decisions support in medical information systems of a medical organization. Medical Doctor and IT. 2017;(2):60–72. (In Russ.).

2. Malygina Yu.S., Orlova K.Yu. Forecast of sales for drug groups in the pharmacy segment of the Russian Federation. In: Problems of economics of modern industrial complexes. Financing and lending in the Russian economy: Methodological and practical aspects: Proceedings of the XV All-Russian Scientific and Practical Conference, 05 December 2022, Samara, Russia. Samara: Samara Scientific Center of the RAS; 2022. P. 16–27. (In Russ.).

3. Fazekas M., Veljanov Z., de Oliveira A.B. Predicting pharmaceutical prices. Advances based on purchase-level data and machine learning. BMC Public Health. 2024;24(1). https://doi.org/10.1186/s12889-024-19171-9

4. Ragb H. Multi-layered deep learning perceptron based model for predicting drug price changes. TechRxiv. URL: https://doi.org/10.36227/techrxiv.24417697.v1 [Accessed 15th November 2025].

5. Puchkov E.V. Comparative analysis of the artificial neural network training algorithms. Engineering Journal of Don. 2013;(4). (In Russ.). URL: https://ivdon.ru/en/magazine/archive/n4y2013/2135

6. Parfenov A.K. Analysis and assessment of investment risks in investing using mathematical modeling. Young Scientist. 2024;(49):9–13. (In Russ.).

7. Filina O.V., Sungatullin K.I. Forecasting the market needs for medicines based on artificial intelligence technologies. Economy and Security. 2025;(1):22–25. (In Russ.).

8. Goryachkin B.S., Chechnev A.A. Sensitivity analysis of binary classification metrics to data imbalance. E-Scio. 2021;(4):23–34. (In Russ.).

9. Nikolenko S., Kadurin A., Arkhangelskaya E. Deep learning. Dive into the world of neural networks. Saint Petersburg: Piter; 2020. 480 p. (In Russ.).

10. Kostyashin N.A., Kolbina O.N., Yagotinceva N.V. Use of automated tools for collecting information on sites. Information Technologies and Systems: Management, Economics, Transport, Law. 2020;(3):11–17. (In Russ.).

11. Stroyakovsky D.L., Abramov M.E., Demidov L.V., et al. Practical recommendations for the drug treatment of skin melanoma. Malignant Tumoursis. 2022;12(3S2-1):287–306. (In Russ.). https://doi.org/10.18027/2224-5057-2022-12-3s2-287-306

12. Laktionov K.K., Artamonova E.V., Breder V.V., et al. Practical recommendations for the drug treatment of non-small cell lung cancer. Malignant Tumoursis. 2021;11(3S2-1):36–54. (In Russ.). https://doi.org/10.18027/2224-5057-2021-11-3s2-02