Accurate medicine quantification is essential in Sub-Saharan Africa (SSA), where supply chain inefficiencies and persistent stock-outs undermine treatment outcomes. Although Artificial Intelligence (AI) and Machine Learning (ML) offer promising solutions, their use in the region remains poorly documented. This systematic review was prospectively registered with PROSPERO (CRD42024552735) and conducted in accordance with PRISMA 2020 guidelines. A comprehensive search of PubMed, Embase, Web of Science, and Google Scholar were searched through May 2024. Eligible studies were peer-reviewed and assessed for technologies aimed at improving quantification of medicines in SSA. Screening was performed using Covidence software and study quality was assessed with Mixed Methods Appraisal Tool (MMAT). Seven studies met the inclusion criteria. Three evaluated AI/ML technologies: Random Forest showed strong predictive accuracy with Root Mean Square Error (RMSE) of Train: 1.137 | Test: 1.23 and R² of Train: 0.78 | Test: 0.71; Long Short-Term Memory (LSTM) showed excellent model fit with RMSE of Train: 2.0 | Test: 2.043 and R² of Train: 0.952 | Test: 0.912. Four evaluated non-AI technologies: SMS for Life, Forced Ordering Max–Min system, and E+TRA Health system, reported reduced stock-outs and improved supply visibility. Most studies were conducted in Rwanda, Tanzania, and Uganda and demonstrated moderate to high methodological quality. AI/ML models demonstrate potential to improve medicine quantification accuracy in SSA, while non-AI digital tools effectively reduce stock-outs. Therefore, wider adoption of context-appropriate solutions is needed to strengthen supply chain in the region.
Reuben et al. (Sat,) studied this question.