The increase in pharmaceutical expenditure in many countries has raised concerns about the sustainability of healthcare services. A key factor that can significantly improve the system is the implementation of tools and policies that more accurately predict drug demand. In this context, a study entitled “A prediction framework for pharmaceutical drug consumption using short time-series,” conducted by Francesco Bertolotti, Fabrizio Schettini, Lucrezia Ferrario, Daniele Bellavia, and Emanuela Foglia, affiliated with the School of Industrial Engineering at LIUC and the Health Care Datascience Lab (HD-LAB), presents an innovative forecasting framework for drug consumption using short time-series. This study addresses the complexity of pharmaceutical demand forecasting by integrating various types of historical data and simulating the generative process of drug consumption. The proposed algorithm generates a distribution of probable values, allowing the use of not only the central forecast value but also its uncertainty, which is crucial for decision-makers in such a critical and complex context.
The results demonstrate that the method enables reliable forecasts, even with short time series, while also ensuring the explainability of results to policymakers. The methodology was validated through back-testing, in which forecasts were compared with actual data. This process yielded an accuracy of over 83%.
This research was supported by EGUALIA and made use of data provided by AIFA and IHS.
This innovative tool can therefore assist in ensuring a drug supply aligned with patients’ healthcare needs, optimising the healthcare supply chain and the utilisation of public resources.
