DOI:
https://doi.org/10.14483/23448350.23604Published:
12/20/2025Issue:
Vol. 52 No. 2 (2025): May-August 2025Section:
Research ArticlesForecasting for Social Welfare: Time Series for Optimizing the Equitable Distribution of High-Cost Medicines
Forecasting para bienestar social: series de tiempo para optimizar la distribución equitativa de medicamentos de alto costo
Keywords:
AI4SG, social welfare, fair distribution, forecasting, high-cost medicines, predictive models, time series (en).Keywords:
AI4SG, bienestar social, distribución justa, forecasting, medicamentos de alto costo, modelos predictivos, series de tiempo (es).Downloads
Abstract (en)
Access to high-cost medication represents one of the most complex and critical challenges in managing health systems, especially in regions with economic and structural limitations. To mitigate said impact, this study proposes the use of forecasting models, integrating time series to anticipate demand and promote a more efficient, equitable, and fair distribution. To this effect, a monthly dataset is used which includes a four-year record of hospital consumption, which has been processed and enriched with public cost information in order to identify the medications with the highest economic impact. The MLOps methodology is employed for the creation, automation, and validation of three models: Prophet, XGBoost, and SARIMAX. Each of these models is optimized through hyperparameter search and evaluated using cross-validation and metrics such as MAPE, RMSE, and MAE. The results show that Prophet stands out in relative accuracy by capturing smooth trends and seasonalities, that XGBoost fits better in absolute units when facing abrupt variations, and that SARIMAX robustly models recurring cycles. These strengths allow adapting the approach according to planning objectives, e.g., optimizing percentage accuracy or reducing volume-related error.
Abstract (es)
El acceso a medicamentos de alto costo representa uno de los desafíos más complejos y críticos en la gestión de los sistemas de salud, especialmente en regiones con limitaciones económicas y estructurales. Con el fin de mitigar dicho impacto, este estudio propone el uso de modelos de forecasting, integrando series de tiempo para anticipar la demanda y promover una distribución más eficiente, equitativa y justa. Para ello, se utiliza un conjunto de datos mensuales que incluye un registro de cuatro años de consumo hospitalario, el cual ha sido tratado y enriquecido con información pública de costos, en aras de identificar los medicamentos de mayor impacto económico. Se utiliza la metodología MLOps para la creación, automatización y validación de tres modelos: Prophet, XGBoost y SARIMAX. Cada uno de estos modelos es optimizado con búsqueda de hiperparámetros y es evaluado mediante validación cruzada y métricas como el MAPE, el RMSE y el MAE. Los resultados muestran que Prophet destaca en precisión relativa, al captar tendencias y estacionalidades suaves; que XGBoost se ajusta mejor en unidades absolutas ante variaciones abruptas; y que SARIMAX modela ciclos recurrentes con solidez. Estas fortalezas permiten adaptar el enfoque según los objetivos de planificación, e.g., optimizar precisión porcentual o reducir el error en volumen.
References
Allwright, S. (2022). RMSE vs MAPE, which is the best regression metric. https://stephenallwright.com/rmse-vs-mape/
Alharbi, F. R., & Csala, D. (2022). A seasonal autoregressive integrated moving average with exogenous factors (SARIMAX) forecasting model-based time series approach. Inventions, 7(4), 94. https://doi.org/10.3390/inventions7040094
Box, G. E., Jenkins, G. M., Reinsel, G. C., & Ljung, G. M. (2015). Time series analysis: Forecasting and control. John Wiley & Sons.
Espinosa-Zúñiga, J. J. (2020). Aplicación de algoritmos Random Forest y XGBoost en una base de solicitudes de tarjetas de crédito. Ingeniería, Investigación y Tecnología, 21(3), 22. https://doi.org/10.22201/fi.25940732e.2020.21.3.022
Chen, T., & Guestrin, C. (2016). XGboost: A scalable tree boosting system. En ACM (Eds.), Proceedings of the 22nd ACM SIGKDD International Conference on Knowledge Discovery and Data Mining (pp. 785-794). ACM.
Hezam, I. M., et al. (2024). Artificial intelligence’s impact on drug delivery in healthcare supply chain management: data, techniques, analysis, and managerial implications. Journal of Big Data, 11(1), 1-25. https://doi.org/10.1186/S40537-024-01049-7
Li, L., Kand, Y., and Li, F. (2023). Bayesian forecast combination using time-varying features. International Journal of Forecasting, 39(3), 1287-1302. https://doi.org/10.1016/j.ijforecast.2022.06.002
MLOps (2023). MLOps Principles. Retrieved from: https://ml-ops.org/content/mlops-principles
Mousa, B. A., and Al-Khateeb, B. (2023). Predicting medicine demand using deep learning techniques: A review. Journal of Intelligent Systems, 32, e0297. https://doi.org/10.1515/jisys-2022-0297
Muraina, I. O., Adesanya, O., & Abam, S. O. (2023). Data analytics evaluation metrics essentials: Measuring model performance in classification and regression [Artículo de conferencia]. Congress on Scientific Researches and Recent Trends 11, Cappadocia, Nevşehir, Türkiye.
Nguyen, A., Lamouri, S., Pellerin, R., Tamayo, S., & Lekens, B. (2022). Data analytics in pharmaceutical supply chains: State of the art, opportunities, and challenges. International Journal of Production Research, 60(22), 6888-6907. https://doi.org/10.1080/00207543.2021.1950937
Petrovic, A., Antonijevic, M., Strumberger, I., Jovanovic, L., Savanovic, N., & Janicijevic, S. (2023). The XGboost approach tuned by TLB metaheuristics for fraud detection [Artículo de conferencia]. 1st International Conference on Innovation in Information Technology and Business (ICIITB 2022).
Rostami-Tabar, B., et al. (2022). Forecasting for social good. International Journal of Forecasting, 38(3), 1245-1257. https://doi.org/10.1016/j.ijforecast.2021.02.010
Seyedan, M., & Mafakheri, F. (2020). Predictive big data analytics for supply chain demand forecasting: methods, applications, and research opportunities. Journal of Big Data 2020, 7(1), 1-22. https://doi.org/10.1186/S40537-020-00329-2
Shahriari, B., Swersky, K., Wang, Z., Adams, R. P., & De Freitas, N. (2015). Taking the human out of the loop: A review of Bayesian optimization. Proceedings of the IEEE, 104(1), 148-175. https://doi.org/10.1109/JPROC.2015.2494218
Shohan, M. J. A., Faruque, M. O., & Foo, S. Y. (2022). Forecasting of electric load using a hybrid LSTM-neural prophet model. Energies, 15(6), 2158. https://doi.org/10.3390/en15062158
Taylor, S. J., & Letham, B. (2018). Forecasting at scale. The American Statistician, 72(1), 37-45. https://doi.org/10.1080/00031305.2017.1380080
Wang, B., Liu, X., Chi. M, & Li, Y. (2024). Bayesian network based probabilistic weighted high-order fuzzy time series forecasting. Expert Systems with Applications, 237, e121430. https://doi.org/10.1016/j.eswa.2023.121430
Wong, A., Flanagan, T., Covington, E. W., Nguyen, E., Linn, D., Brummel, G., Hoffmaster, B., Isaacs, D., & Kane-Gill, S. L. (2025). Forecasting the impact of artificial intelligence on clinical pharmacy practice. Journal of the American College of Clinical Pharmacy, 8(4), 302-310. https://doi.org/10.1002/jac5.70004
Zhu, X., Ninh, A., Zhao, H., & Liu, Z. (2021). Demand forecasting with supply-chain information and machine learning: Evidence in the pharmaceutical industry. Production and Operations Management, 30(9), 3231-3252. https://doi.org/10.1111/poms.13426
How to Cite
APA
ACM
ACS
ABNT
Chicago
Harvard
IEEE
MLA
Turabian
Vancouver
Download Citation
License
Copyright (c) 2025 Donovan-Steven Picón-Sossa, Marco-Javier Suarez Barón, Juan-Sebastián González-Sanabria
This work is licensed under a Creative Commons Attribution-NonCommercial-ShareAlike 4.0 International License.
When submitting their article to the Scientific Journal, the author(s) certifies that their manuscript has not been, nor will it be, presented or published in any other scientific journal.
Within the editorial policies established for the Scientific Journal, costs are not established at any stage of the editorial process, the submission of articles, the editing, publication and subsequent downloading of the contents is free of charge, since the journal is a non-profit academic publication. profit.
