Abstract:

The growing prevalence of diet-related diseases such as obesity, diabetes, and cardiovascular disorders is closely linked to the excessive consumption of unhealthy food products and overeating. Identifying harmful food items based on their chemical and nutritional composition has therefore become an important research direction in public health and artificial intelligence (AI). This study presents an AI-based approach to recognizing unhealthy food products by analyzing their ingredient lists and nutritional profiles. The dataset used in the research was collected from various online sources, including publicly available food databases and manufacturer websites. Several machine learning models were trained and evaluated to classify food products according to their health impact. The algorithms included linear regression, ridge regression, logistic regression, stochastic gradient descent (SGD), multilayer perceptron (MLP) neural networks, passive-aggressive classifier, support vector machines (SVC) with Gaussian and linear kernels, k-nearest neighbors, naive Bayes, decision trees, random forests, gradient boosting, extremely randomized trees (Extra Trees), AdaBoost, and neural network classifiers. Experimental results demonstrated that ensemble and neural-based methods achieved the highest classification accuracy, confirming the effectiveness of AI in supporting nutritional assessment and promoting healthier dietary choices.