Abstract
Relevance. Population growth in urban agglomerations creates a quantifiable gap between the demand for medical care and the availability of personnel, medicines, financial resources, and logistical support. The Almaty agglomeration is home to approximately 3.5 million residents, with a projected population of 4.5 million by 2030. According to the Ministry of Health of the Republic of Kazakhstan, in 2023, the shortage of physicians amounted to 4,864 positions, and the concurrent employment rate reached 1.4, reflecting systemic staff overload. Objective. To analyze current models for forecasting healthcare resource needs in urban agglomerations, with a focus on machine learning tools, and to assess their applicability to the conditions of the Almaty agglomeration. Materials and methods: A literature search was conducted in the PubMed, Scopus, Web of Science Core Collection, and WHO IRIS databases for the period from January, 2010 to March, 2026. The review included original studies, systematic reviews, and WHO/OECD methodological reports containing quantitative predictive models with accuracy metrics (MAE, RMSE, MAPE, AUROC, R²) at the level of urban agglomerations or regions with an urbanization rate ≥ 50%. Of the 1,286 records identified, 57 publications were included in the review. Results. Statistical models were the most common (40.4%), followed by supervised machine learning methods (34.0%), hybrid models (17.0%), and unsupervised machine learning methods (4 publications, or 8.5%). Only 3 validated models (6.4%) used data from Kazakhstan and cover the human resources category exclusively. No model covered all four resource categories simultaneously. Conclusions. The methodological findings of this review justify the development of an integrated forecasting model based on unsupervised machine learning for the Almaty metropolitan area, covering at least three resource categories and a 5-10-year forecast horizon in accordance with national planning documents.