Chronic Obstructive Pulmonary Disease (COPD) is a chronic respiratory condition that significantly impacts the health and quality of life of millions globally. Immediate identification is essential for enhancing patient outcomes; nevertheless, conventional diagnostic techniques frequently necessitate intricate and costly apparatus. To mitigate this deficiency, we present an energy-efficient deep learning model for COPD detection that utilises respiratory sound analysis. This method utilises deep learning to autonomously classify COPD based on audio recordings of respiratory sounds, rendering it suitable for real-time, mobile, and embedded medical systems where power efficiency is paramount. The suggested model uses a lightweight 1D convolutional neural network (CNN), optimised for energy economy, to extract significant characteristics from raw respiratory sound signals. Methods such as model pruning, quantisation, and early exit strategies are employed to diminish computing demands and energy usage during inference while maintaining accuracy. These optimisations guarantee that the model functions efficiently in resource-limited situations, such as wearable health devices or smartphones that are frequently employed for the continuous monitoring of respiratory health. Experiments demonstrate that the optimised deep learning model achieves elevated classification accuracy for COPD detection while markedly decreasing energy usage relative to conventional methods. This makes it a practical option for mobile healthcare and remote patient monitoring, allowing for quick COPD diagnosis in both advanced and less developed areas. This work highlights the promise of lightweight deep learning methodologies in enhancing energy-efficient medical devices for the management of respiratory diseases.