BACKGROUND: The healthcare industry is rapidly evolving due to digital innovation and technological advancements. The increasing volume of healthcare data necessitates efficient analytical methods to extract meaningful insights. Traditional health data analysis platforms primarily focus on data collection, aggregation, processing, analysis, visualisation, and interpretation. However, challenges remain in optimising these processes for effective disease prediction and decision-making. METHODS: This study proposes the k-means termite clustering model (KTCM) as a novel optimisation approach for healthcare data analysis. The model integrates graph reduction techniques for data preprocessing, followed by storage in a clinical database. A mining algorithm is employed to analyse the processed data, enhancing predictive accuracy. Healthcare professionals receive training on standardised prediction methodologies to refine disease forecasting based on historical benchmarks. The model's performance is evaluated using statistical metrics, including R², REMS, MSE, MAE and MAPE. RESULTS: The proposed KTCM model demonstrates superior predictive performance, achieving an R² value of 99.7%, surpassing other existing methods. The advanced clustering and optimisation techniques improve the accuracy and efficiency of disease prediction, thereby aiding healthcare professionals in making informed decisions. CONCLUSION: The KTCM approach significantly enhances healthcare data analysis by optimising disease prediction through efficient clustering and mining techniques. The model's high accuracy and improved parameter optimisation validate its effectiveness in clinical decision support. Future work may explore further refinements in algorithmic performance and real-time implementation in healthcare systems.