Predicting Adolescent Obesity with DeepHealthNet: A Deep Learning Approach

The increasing prevalence of adolescent obesity is a significant public health concern due to its association with chronic diseases and long-term health risks. Leveraging artificial intelligence (AI) to predict obesity risk and provide personalized health insights has emerged as a promising solution. This study introduces DeepHealthNet, a deep learning-based obesity prediction system designed to offer individualized feedback to adolescents based on key health indicators such as height, weight, waist circumference, calorie intake, and physical activity levels.To develop and evaluate the proposed model, health data were collected from 321 adolescents using the Would You Do It! (WUDI!) application. The DeepHealthNet framework employs data augmentation techniques to enhance learning, even with limited daily health records, resulting in improved prediction accuracy (overall accuracy: 88.42%). Notably, performance variations were observed between male (93.20%) and female (91.63%) participants, highlighting potential gender-based differences in obesity prediction. Additionally, statistical analysis (p < 0.001) demonstrated that DeepHealthNet significantly outperforms traditional models in obesity classification.The findings suggest that the proposed system can serve as an effective tool for early detection and prevention of adolescent obesity, enabling timely interventions and promoting healthier lifestyle choices.