The task focuses on developing smaller, more efficient AI models to reduce energy consumption and costs while maintaining high performance. Leveraging techniques like distributed training, quantization, and model redesign, we aim to create sustainable AI solutions tailored for geriatric care. By utilizing Human-Centric Benchmarking datasets, we ensure these models remain effective in real-world applications. Our work emphasizes system-level performance, exploring how resource-efficient components can maintain robustness within feedback-driven systems. This study will largely support the deployment of AI in healthcare while aligning with global sustainability goals.