Optimizing Diet Personalization with Hybrid Clustering and Deep Learning Methods

Abstract

Personalized nutrition is an emerging domain that aims to tailor dietary recommendations according to individual characteristics such as genetics, lifestyle, metabolic markers, and health goals. Traditional approaches often fall short in capturing the inherent complexity and variability of human dietary needs. This research proposes an integrated framework combining hybrid clustering and deep learning techniques to enhance diet personalization. We employ a two-step hybrid clustering strategy involving K-means and hierarchical clustering to group individuals with similar nutritional profiles. Subsequently, a deep learning model, specifically a multi-layer perceptron (MLP), is trained to predict optimized diet plans tailored to these clusters. Extensive experiments on real-world nutrition datasets demonstrate that our hybrid method outperforms traditional clustering and standalone machine learning models in terms of personalization accuracy, dietary adherence, and health outcome improvement. The proposed framework provides a robust, scalable, and intelligent pathway toward achieving highly individualized dietary recommendations.