Effective self-management of diabetes is vital to prevent severe complications and reduce the burden on healthcare systems. The study explores the integration of knowledge graph (KG) and natural language generation (NLG) to enhance personalised health recommendations within a digital healthcare system, addressing the limitations in real-time health data analysis and care plan personalisation. The methodology involves constructing KGs using Neo4j to integrate data from glucose monitoring devices, dietary logs, and physical activities, paired with OpenAI’s GPT-3.5 for NLG to generate personalised insights. Results show the system’s ability to uncover health patterns and deliver actionable recommendations, allowing users to visualise relationships between health data points for better management. Future work will focus on expanding capabilities, adding data sources, and improving the user interface and security for a more comprehensive and secure personalised healthcare system.