@unpublished{osei2026farmdirect,
  author = {Osei, H. A.},
  title = {FarmDirect: Data-Driven Pricing and Risk-Adjusted Allocation for Ghanaian Tomato Markets},
  note = {Submitted to arXiv; on hold pending moderation (cross-disciplinary classification)},
  doi = {},
  file = {FarmDirect_Research.pdf}
}

This study presents four connected empirical components, price prediction, market clustering, risk-adjusted allocation, and an agent-based simulation of farmer selling behavior, built on the 2025 MoFA/SRID government price data for Ghanaian tomato markets. Observable features such as region, market, and time of season predict prices well enough to beat a naive last-known-price baseline, though the useful signal is overwhelmingly seasonal. Markets cluster into distinct price-behavior regimes in which price level and price volatility are largely independent properties, a finding that is corroborated independently by the price-prediction model’s own residual errors. Building on these two results, a Markowitz-style risk-adjusted allocation model concentrates supply on a small number of markets that jointly balance expected revenue against measured risk. Finally, an agent-based simulation tests whether eliminating a middleman actually benefits farmers, finding that direct selling outperforms selling through a middleman once farmers have only modest, better-than-random market information, though this benefit is not distributed evenly across market types: farmers selling into more volatile markets bear substantially higher revenue variance even under perfect information. The analysis links prediction, clustering, allocation, and simulation into a connected analytical pipeline.

@unpublished{osei2026wastelearning,
  author = {Osei, H. A. and Osei, C.T. and Asobayire, D.Y.},
  title = {Learning from waste: Machine Learning for health risk prediction and computer vision-based sorting in Ghana},
  note = {arXiv:2608.25759 [cs.LG], cross-listed cs.CV, cs.CY},
  doi = {2608.25759},
  file = {ATONSU_WASTE_ML_PROJECT.pdf}
}

The inappropriate disposal of solid waste remains a significant public health and environmental concern worldwide, including in Ghana. Poor sanitation and improper waste management practices contribute to substantial economic costs and avoidable deaths annually. In 2022, a field study in Atonsu, Kumasi, Ghana, reported a community-perceived relationship between household waste disposal and illness patterns, but only through descriptive analysis without quantitative validation. This study extends that investigation using two data-driven approaches. First, a Random Forest classifier was developed to predict illness categories using waste disposal practices and demographic survey data. On a held-out group of respondents who reported illness (N=69), the model obtained a macro F1 score of 0.63, with disposal method emerging as the most important substantive predictor of illness type. Second, a MobileNetV2 image classification model enabled automated waste sorting via visual recognition, achieving 88.2% accuracy and a macro F1 score of 0.87 on the test set (N=415). The vision-based approach offers an affordable, camera-driven alternative to complex multi-sensor systems, making it highly suitable for resource-constrained settings. Taken together, the findings provide quantitative evidence for a community health relationship previously documented only qualitatively. They demonstrate the potential for automated waste-sorting in low-resource environments. Importantly, the results illustrate that technological performance alone does not guarantee public health improvements; effective institutional support and implementation are equally necessary.

@unpublished{osei2026nosocomial,
  author = {Osei, C. T. and Asare, A. O. and Oti-Agyen, Y. and Osei, H. A. and Amooba, P.},
  title = {Knowledge and Compliance with Standard Precautions for Nosocomial Infection Prevention Among Undergraduate Nursing and Midwifery Students in Ghana},
  note = {medRxiv 2026.07.07.26357431},
  doi = {10.64898/2026.07.07.26357431},
  file = {2026.07.07.26357431v1.full.pdf}
}

Background: Healthcare-associated infections represent a significant patient safety challenge in sub-Saharan Africa, where gaps in infection prevention and control practice remain prevalent. Nursing and midwifery students are particularly vulnerable during clinical training, yet evidence on their IPC knowledge and compliance in Ghana is limited. Objective: To assess the knowledge of nosocomial infections and the degree of compliance with standard precautions among third-year nursing and midwifery students at Kwame Nkrumah University of Science and Technology (KNUST), Ghana. Methods: A descriptive cross-sectional study was conducted between 28 June and 9 July 2021 at KNUST, Kumasi, Ghana. A total of 150 third-year students (65 nursing, 85 midwifery) completed a structured, self-administered questionnaire adapted from WHO and CDC guidelines. Knowledge was assessed using a 19-item binary scale and compliance using a 17-item Likert-type scale. Chi-square tests, Fisher’s exact test, and Spearman’s rank correlation were used to examine associations between knowledge and compliance. Results: Overall, 143 respondents (95.3%) were categorised as having high knowledge (mean score: 16.44/19), and 112 (74.7%) reported high compliance with standard precautions (mean score: 59.13/68). Compliance was strongest for hand hygiene and glove use but notably lower for PPE use during splash-risk procedures and safe needle-handling practices. No statistically significant association was found between categorised knowledge and compliance levels (χ^2 = 0.47, p = 0.491), though a modest positive correlation emerged when analysed as continuous scores (Spearman’s ρ= 0.326, p < 0.001). Conclusion: High knowledge of nosocomial infections did not translate uniformly into high compliance across all standard precaution domains, pointing to a need for practical training, simulation-based learning, and supervised clinical reinforcement in nursing and midwifery education in Ghana.