MULTI-OBJECTIVE OPTIMIZATION MODELS FOR PREDICTING DISEASE PROGRESSION IN BIO-MATHEMATICAL SYSTEMS

The escalating global reliance on predictive analytics in healthcare highlights the urgent need for efficient, adaptable models, particularly in resource-constrained regions like Ghana. Despite a global 48% rise in predictive model adoption between 2020 and 2024, Ghana's uptake remains low at 6%, risking healthcare inefficiencies and delayed interventions. This study developed multi-objective optimization models, using hybrid Genetic Algorithms and Particle Swarm Optimization, to enhance disease progression prediction across Ghanaian healthcare systems. Utilizing secondary data from 105 respondents and applying Pearson correlation and multiple regression analyses, the findings revealed that algorithmic efficiency (β = 0.391, p < 0.001), objective function formulation (β = 0.352, p < 0.001), and model adaptability (β = 0.287, p < 0.001) significantly improved disease prediction accuracy, achieving an average sensitivity of 86.4%, specificity of 83.9%, and computational time of 9.3 seconds per prediction, with an overall model explanatory power of R² = 0.672. The results demonstrate that integrating robust multi-objective optimization strategies can simultaneously enhance sensitivity, specificity, and computational efficiency, addressing critical predictive gaps in low-resource settings. These findings imply the necessity for Ghanaian health policymakers to invest in computational infrastructure, prioritize hybrid optimization models, and integrate real-time adaptability into clinical workflows. Future research should focus on expanding real-time learning capabilities and incorporating broader genomic data to bolster predictive robustness.