AUTHOR=Venkata Krishna P. , Saritha V. , Pachipulusu Pratap , Aruna M. , Jayasri P. 
  
TITLE=Blockchain-enabled real-time personalized nutrition recommendation framework using IoT and AI
  
JOURNAL=Frontiers in Blockchain
  
VOLUME=Volume 9 - 2026
  
YEAR=2026
  
URL=https://www.frontiersin.org/journals/blockchain/articles/10.3389/fbloc.2026.1765645
  
DOI=10.3389/fbloc.2026.1765645
  
ISSN=2624-7852
  
ABSTRACT=Today, people need personalized diet plan based on their health conditions. Latest technologies like Internet of Things (IoT), federated Learning, blockchain technology, and wearable devices help in gathering the information required to recommend personalized and nutritional diet plan and also maintain the data securely. As most of the existing system that recommend nutrition diet has many limitations like lack of privacy, less user engagement, usage of AI models that are not transparent and centralized data storage. Hence, Blockchain enabled Real-Time Personalized Health and Nutrition Management (BRPHM) framework is proposed in this paper. BRPHM is a multi-layer architecture which includes IoT data acquisition layer, Blockchain data management layer, federated AI processing layer, and a recommendation layer. BRPHM introduces a new parameter called Personalized Health Nutrition Index (PHNI) based on which recommendations are given to the user. A weighted health model based on environmental, nutrition, activity, physiological features determine PHNI value. The performance of the proposed framework is evaluated in terms of accuracy, recall, precision, F1-score, Mean Absolute Error (MAE), Root Mean Squared Error (RMSE), latency, system availability, privacy score, and scalability score and is compared with ESCIFS, SEDCAM-2E and PNBDF. The results indicate that the proposed framework, BRPHM enhances the performance by 8%–22% in terms of classification metrics (accuracy, precision, recall and F1-score), 33%–53% in terms of forecasting metrics (MAE and RMSE), 42%–59% in terms of latency, 1.4%–2.8% in terms of system availability, 10%–27% in terms of privacy score and scalability score when compared to ESCIFS, SEDCAM-2E and PNBDF. The results also projects PHNI correlation score and Micro-action engagement score which indicates that the model is accurate and the system is effective.