Obesity, a complex multifactorial condition, poses significant global health challenges, necessitating precise prediction and management strategies. This study explores the integration of DNA methylation data into Body Mass Index (BMI) prediction models using advanced computational techniques. The primary aim is to enhance the precision of BMI predictions while uncovering novel insights into the epigenetic factors associated with obesity-related health risks. Through the utilization of a Deep Feedforward Neural Network (DFNN) model and backward feature selection method, DNA methylation patterns strongly associated with BMI are identified. The DFNN model exhibits moderate performance in BMI prediction, with R-squared values ranging from 0.36 to 0.77 across training, validation, and test datasets. Backward feature selection further refines the predictive accuracy, revealing an average R-squared value of 0.41 and highlighting the effectiveness of this approach in identifying relevant DNA methylation patterns. Fisher’s exact test uncovers significant associations between gene sets, elucidating potential regulatory relationships in obesity-related pathways. Gene ontology enrichment analysis underscores the involvement of lipid metabolism and cholesterol regulation in obesity pathogenesis. Ethical considerations regarding participant confidentiality and data privacy are meticulously addressed, ensuring adherence to ethical standards throughout the research process. In conclusion, this study underscores the promise of integrating DNA methylation data into BMI prediction models, offering novel avenues for personalized medicine and advancing our understanding of obesity-related health outcomes.