Classificação de Níveis de Obesidade através de Redes Neurais Multicamadas: Uma Abordagem Baseada em Hábitos de Vida e Dados Antropométricos

Abstract

Obesity and the risk of cardiovascular disease represent growing challenges for global public health. Preventive diagnosis and accurate classification of risk levels are fundamental for effective medical interventions. This study presents an artificial intelligence-based approach, using a multilayer perceptron neural network to classify the obesity level of individuals based on their eating habits and physical condition. The proposed model, structured with dense layers, ReLU activation function, dropout regularization, and Adam optimizer, achieved an overall accuracy of 94\% on the test data with consistent generalization. Metric analysis demonstrated remarkable robustness, achieving an F1-Score of 1.00 for the Grade III Obesity class, with zero false negatives in the highest risk category.