Diabetes-associated cognitive decline represents an emerging public health burden, particularly in low-resource settings where social determinants and genetic predisposition remain under-integrated in predictive modeling. This study proposes a novel Hybrid Socio-Genomic Risk Integration Algorithm (HSGRIA) that combines clinical biomarkers, food security indices, psychosocial stress metrics, and genomic susceptibility markers to improve early risk prediction of cognitive decline among individuals with type 2 diabetes. Data were derived from outpatient cohorts in Kumasi, Ghana, incorporating fasting blood glucose, lipid profiles, anthropometric measures, and genotypic data including APOE-associated variants. Psychosocial and dietary variables were quantified using validated survey instruments. The proposed model integrates multivariate statistical weighting with machine learning-assisted feature selection to generate individualized risk scores. HSGRIA was evaluated against five established approaches including logistic regression, random forest, support vector machines, gradient boosting, and traditional clinical risk scoring systems. Performance assessment using cross-validation demonstrated that HSGRIA achieved superior predictive accuracy, improved sensitivity in early-stage detection, and greater robustness to missing socio-clinical data. Notably, the integration of food insecurity and stress-related variables significantly enhanced model performance, highlighting the importance of non-biological determinants in neurodegenerative risk pathways. This work provides a scalable and interpretable framework for precision risk stratification in underserved populations and supports the integration of socio-environmental and genomic data in neurological disease prediction.