Predicting Obesity with Deep Learning: Enhancing Accuracy through Data Preprocessing

The goal of this work is to improve adult obesity prediction via the use of a Multilayer Perceptron (MLP) model with sophisticated data preparation methods. The lack of accuracy in the prediction of obesity necessitates deep learning approaches in this work, where a Multilayer Perceptron (MLP) will be used. For this purpose, the model echo high levels of data pretreatment mechanisms including data cleaning, normalization, feature engineering and dimensionality reduction to ensure the provision of high quality inputs. The suggested model outperformed the previous approaches; it provided an accuracy of 93.8%; precision of 92.7, F1 score of 93.2 and ROC AUC score of 95.1. These results suggest that the model itself might give higher prognosis, which will be also helpful in health promotion and individual treatments for avoiding obesity. This paper concludes with a review of the significance of these discoveries for healthcare applications.