APPLICATIONS OF MACHINE LEARNING-BASED OPTIMIZATION IN BIO-MATHEMATICAL MODELING OF DISEASE DYNAMICS

The growing complexity of infectious disease dynamics in Ghana necessitates innovative predictive tools, particularly as traditional models have struggled with real-time accuracy. Capturing the urgency of over 300 global epidemics recorded annually between 2020 and 2024 (WHO, 2024), this study justifies the critical need for machine learning-based optimization in enhancing bio-mathematical disease models. The research aimed to assess how supervised, unsupervised, and reinforcement learning techniques, alongside external epidemiological factors, could improve predictive accuracy, computational efficiency, generalizability, and model stability. Using a quantitative explanatory design, secondary data from 105 validated datasets spanning Greater Accra, Ashanti, and Eastern regions were analyzed through descriptive statistics, correlation, and regression modeling. Findings revealed high predictive gains, with machine learning models achieving an average predictive accuracy of 89.8% by 2024 and computational efficiency improving by 26.8%, while the correlation between optimization techniques and model performance reached a strong r = 0.865. Regression analysis confirmed that supervised learning (β = 0.472, p < 0.001) had the most significant impact. The study concludes that integrating machine learning optimization dramatically boosts Ghana’s disease modeling capabilities, offering transformative potential for early interventions and resource management. Implications highlight the urgent need for national adoption of AI-driven disease forecasting, with recommendations including investment in real-time surveillance systems, training of local health professionals in AI methods, and enhanced integration of climate and mobility data to refine predictive modeling frameworks.