Background and Rationale: Chronic diseases such as type 2 diabetes, cardiovascular disease, and hypertension are major global health burdens, with dietary behaviour being a key modifiable risk factor. While previous studies have established associations between specific nutrients and disease outcomes, there is limited research on using comprehensive dietary behaviour patterns for disease risk prediction and personalised intervention.
Research Questions: (1) Can machine learning models accurately predict chronic disease risk based on dietary behaviour patterns? (2) What are the most predictive dietary features for different chronic diseases? (3) How can these predictive models inform personalised dietary recommendations?
Objectives: This project aims to: (1) Develop and validate machine learning models (including gradient boosting, random forest, and deep learning approaches) to predict the risk of type 2 diabetes, cardiovascular disease, and hypertension using dietary intake data; (2) Identify key dietary behaviour patterns and features most predictive of chronic disease risk; (3) Build a personalised dietary recommendation system based on individual risk profiles.
Methods: We will utilise dietary assessment data (24-hour recall, food frequency questionnaires), combined with health outcomes, biomarkers, and demographic variables. Multiple machine learning algorithms will be trained and validated using cross-validation approaches. Feature importance analysis will identify critical dietary factors, which will inform the development of a recommendation framework.
Expected Outcomes: This research will provide validated risk prediction tools and evidence-based dietary recommendations that could support public health interventions and clinical decision-making for chronic disease prevention.
