Leveraging Machine Learning to Predict COVID-19 Vaccination Adoption among Healthcare Professionals in Somalia: A Comparative Analysis

This study explores the application of Machine Learning (ML) techniques to predict COVID-19 vaccination adoption
among healthcare professionals in Somalia. Recognizing the pivotal role of vaccination in controlling the pandemic, and the unique challenges faced by Somalia’s healthcare system, this research aims to identify critical factors influencing vaccine uptake and to develop robust predictive models. We employed a rich dataset comprising demographic, health, and institutional variables gathered from a survey of healthcare workers across various Somali states. Three ML algorithms were evaluated: Logistic Regression, Random Forest, and Gradient Boosting. Each model was rigorously assessed using precision, recall, F1-score, and accuracy metrics. Gradient Boosting emerged as the most effective model, demonstrating the highest accuracy (approximately 85%) and F1 score (about 82%). Logistic Regression provided a baseline for comparison, while Random Forest showed notable strengths in certain aspects, particularly in feature importance analysis. The study also involved a detailed
examination of confusion matrices for each model, revealing specific strengths and weaknesses in predictive capabilities. These matrices provided insights into the models’ performance, particularly in distinguishing between different vaccination behavior categories. Our findings suggest that ML can be a powerful tool in predicting vaccine adoption, with significant implications for public health strategies. The Gradient Boosting model, in particular, shows promise for practical application in designing targeted interventions to improve vaccination rates among healthcare workers. This research contributes to the growing body of knowledge in public health informatics, offering a novel approach to tackling vaccine hesitancy and enhancing pandemic response efforts in Somalia and similar settings.