The thesis explores using knowledge graphs to enhance patients' electronic medical records (EMRs) for predicting hospitalization rates, which currently affect 12.7 million patients. The study highlights challenges in managing comorbidities and proposes a decision support tool, 'Health Predict,' to assist general practitioners in anticipating hospitalizations through the analysis of risk factors. Experimental results demonstrate improved prediction accuracy using advanced data representations and machine learning algorithms, emphasizing the integration of ontological insights from medical knowledge bases.