Leveraging Machine Learning for Personalized Dietary Recommendations, Nutritional Patterns, and Health Outcome Predictions
Year: 2024
Type: article
Abstract: Unhealthy dietary patterns are key contributors to chronic diseases such as obesity, diabetes, and cardiovascular conditions. This study employs machine learning (ML) techniques to analyze dietary intake, identify patterns, and assess their relationships with health outcomes. The aim is to provide personalized dietary recommendations and insights to promote healthier eating habits. Data for this research were sourced from a Kaggle dataset on foods and nutrients and the National Health and Nutrition Examination Survey (NHANES) on health outcomes. Preprocessing steps included data cleaning, feature selection, and transformation using one-hot encoding and scaling techniques. Machine learning algorithms were applied to build a food recommendation system and a diet health check system. Visualizations such as correlation heatmaps, scatter plots, and dashboards further illustrated the relationships between demographic factors, nutrient intake, and health outcomes. The food recommendation system effectively identified foods with similar nutritional profiles to user preferences. For instance, it suggested nutrient-rich alternatives like rice noodles and kale, achieving similarity scores above 0.99 in multiple test cases. The diet health check system analyzed nutrient intake against predefined thresholds and provided tailored recommendations. Excessive carbohydrate, protein, fat, and cholesterol consumption were linked to conditions such as diabetes, coronary heart disease, and cancer, with specific dietary adjustments suggested for improvement. This study demonstrates the power of machine learning in personalizing dietary advice and enhancing health outcomes. By leveraging advanced algorithms and diverse datasets, the developed systems present a scalable solution for promoting balanced diets and mitigating chronic disease risks. Further refinement and broader implementation of these tools are recommended to maximize their impact on public health.
Language: English
Subfield: Health Information Management
Field: Health Professions
Domain: Health Sciences
Open Access status: hybrid
OpenAlex ID: W4409746253