| Medical oxygen is essential for treating childhood pneumonia - the leading cause of death in children under five worldwide. Unfortunately, oxygen is not widely available in many low-resource settings due to challenges such as cost, supply logistics, variability in oxygen demand, poor electricity supply, lack of trained staff, and inadequate maintenance capacity. Currently, no systematic approaches exist to help health systems plan appropriate oxygen supply systems that cost-effectively meet the needs of health facilities given these complex challenges.;This thesis presents a novel operations research (OR) approach to medical oxygen technology planning in low-resource settings. A decision-support model -- OxOpt -- was developed to identify the optimal combination of technologies that will meet the needs of health facilities facing resource constraints. The model applies a unique combination of simulation and optimization approaches never before combined to address a global health resource allocation problem. A discrete-event simulation model simulates health facility-level activities such as clinical demand for oxygen, power interruptions and equipment breakdowns, using primary and secondary data collected specifically for this context. A genetic algorithm-based optimization model identifies optimal technology strategies that will satisfy simulated oxygen needs. The model takes into account seasonal variability in oxygen demand, alternative energy options, and the costs of technology, energy and training. It also estimates the number of patients treated and lives saved as a result of recommended solutions.;The design of the OxOpt model is informed by several important analyses. A study of oxygen concentrator maintenance histories contributes previously lacking data on long-term functionality in a low-resource setting, providing insight into the cost of parts, expected lifespans, frequency of failures and technician training needs. Based on an analysis of alternative energy sources, a decision-tree was developed to guide decisions about appropriate energy choices given system costs and grid electricity availability.;OxOpt functionality is demonstrated for the case of The Gambia, through scenarios exploring healthcare equity in resource allocation, climate change, and changes in technology. An OR approach to oxygen planning, which can be applied more broadly to other health technologies, can improve the efficient use of limited resources and identify cost-effective solutions that save lives. |
