5th International Conference on Primary Care and Global Healthcare

Michael Asante Ofosu is a PhD student in Statistics and Data Science at Auburn University, where he holds a Graduate Teaching Assistantship in the Department of Mathematics and Statistics. He holds a BSc in Statistics from the Kwame Nkrumah University of Science and Technology (KNUST). His research interests include interpretable machine learning, responsible AI, health informatics, and applied biostatistics, with a particular focus on graph neural networks and trustworthy machine learning for high-stakes decision systems, as well as Rashomon-set algorithms for model multiplicity. Prior to joining Auburn, he served as a Senior Data Analyst at Touton SA Ghana, where he built forecasting and analytics pipelines for supply-chain operations. He has authored or coauthored several peer-reviewed publications spanning epidemiological forecasting, health informatics, and applied statistics, including work published in Applied Medical Informatics, Frontiers and IEEE Open Journal of the Computer Society, and currently serves as a peer reviewer for the Majestic International Journal of AI Innovations (MIJAI).

Scabies, a World Health Organization designated neglected tropical disease, affects more than 200 million people worldwide at any given time (World Health Organization, 2023), yet primary care systems across sub-Saharan Africa rarely have access to quantitative tools for anticipating its seasonal burden. This study developed and validated a time-series forecasting framework for monthly scabies incidence at a district hospital in Ghana's Ashanti Region, translating routinely collected surveillance data into actionable lead time for primary care resource planning. Eighty-nine monthly case counts (January 2016 to May 2023) from the Kokofu Hospital Information System, in Ghana's Ashanti Region, were analyzed. Statistical tests confirmed a significant upward trend over time along with a recurring seasonal pattern tied to the dry season. A seasonal time series model was fitted on part of the data and then tested on the remaining, unseen data to check how well it could predict future cases. A complementary regression model was also used to identify which months showed significantly higher or lower case numbers. The model fit the historical case data well and performed reliably when tested against real outcomes it had not seen before. Scabies cases rose sharply during the rainy months of June and July and dropped in November and December each year. Looking ahead, the model predicts that this pattern will continue, with case numbers expected to keep rising overall and to peak again around September. These models offer a low-cost, reliable early-warning signal that can be built into routine clinic records, helping health teams pre-position supplies, plan staffing, and time screening campaigns ahead of the seasonal surge, shifting scabies control from reactive to anticipatory.