A Simulation-Driven Approach for Smart Energy Efficiency Alerts in LLM-Enabled Nutrition Apps

As Large Language Models (LLMs) expand in nutrition apps, concerns about energy use are increasing, particularly for users with limited resources or environmental priorities. This study introduces a framework for real-time energy-efficiency alerts, built on the Star Rating Evaluation Model (SREM), to monitor usage during tasks such as meal suggestions, dietary analysis, and chatbot consultations. Tests were conducted on 50 anonymized nutrition apps with LLM features enabled and disabled, using simulated interaction data to mirror typical patterns. Results showed that LLM functions increased power consumption by about 25% on average. These findings underscore the need for real-time alerts in energy-intensive nutrition apps, and the paper concludes with design guidance and a roadmap for practical validation.