"background": "District hospital systems in Kenya face persistent operational risks, including stock-outs of essential medicines and medical supplies, which compromise service delivery. Existing risk assessment frameworks often lack predictive capacity, relying on retrospective analyses that limit proactive intervention. ", "purpose and objectives": "This study aimed to develop and methodologically evaluate a time-series forecasting model designed to predict critical resource shortfalls, with the objective of providing a tool for pre-emptive risk reduction in hospital supply chains. ", "methodology": "We conducted an intervention study implementing a forecasting model across a network of district facilities. The core model was an autoregressive integrated moving average (ARIMA) formulation: Xt = \ + \1 X{t-1 +. . . + \ Xt-p + \ + \1 -1 +. . . + \ -q, where Xₜ is the commodity stock level at time t. Model performance was evaluated using rolling-origin forecast evaluations, with uncertainty quantified via 95% prediction intervals. ", "findings": "The model demonstrated significant predictive utility, reducing unanticipated stock-out events for five key antiretroviral medicines by an average of 32% (95% CI: 24% to 40%) in intervention hospitals compared to control sites. Forecast accuracy, measured by mean absolute scaled error, was superior to routine replenishment methods. ", "conclusion": "The implemented time-series forecasting model provides a statistically robust methodological tool for anticipating supply chain disruptions, enabling more resilient hospital system management. ", "recommendations": "Health system managers should integrate predictive analytics into routine supply chain monitoring. Further research should focus on adapting the model for integrated disease surveillance and incorporating climate and transport delay variables. ", "key words": "health systems resilience, predictive analytics, supply chain management, ARIMA modelling, sub-Saharan Africa", "contribution statement": "This paper presents a novel application of ARIMA time-series forecasting for proactive health system risk
Achieng et al. (Wed,) studied this question.