The manuscript should contain a self-contained abstract of up to 300 words without citations. It must succinctly present the research's purpose, methodology, With the dawn of smart healthcare, the volume of biomedical data available through wearable sensors, imaging techniques, and electronic health records has been growing exponentially, causing growing computational and energy pressures that jeopardize the sustainability and scalability of medical systems. To deal with this issue, this paper presents a deep learning-based biomedical informatics system that seeks to maximize diagnostic performance and reduce the computational and energy costs in clinical settings. The framework combines both hybrid CNN-LSTM models of temporal biomedical signal analysis models and Transformer-based models of multimodal representation learning models, complemented with adaptive pruning and energy-sensitive scheduling on heterogenous datasets of ECG, EEG, and imaging streams. Also, contrary to the traditional healthcare-based AI methods that prioritize precision without considering sustainability, the given system takes adaptative controls into consideration and continuously adjusts computation and resource distribution under different workload and patient-monitoring conditions, which enhances efficiency and resilience. The mathematical modeling and optimization of the framework is the system performance, measured by classification accuracy, diagnostic latency, energy efficiency, and a sustainability index. Comparative experiments indicate that the suggested method obtains the 5.3 percent enhancement in diagnostic accuracy, the 29.7 percent energy usage decrease, and the 21.3 percent latency reduction relative to the baselines, including Random Forest and regular LSTM models. Moreover, stress tests with peak workloads of patient monitoring verify that the framework maintains high levels of adaptability, whereas traditional models deteriorate considerably. The proposed system is the first to note the transformative nature of sustainable AI in healthcare because it facilitates precise, energy-efficient, and scalable biomedical decision support. The current study makes energy-conscious biomedical informatics one of the pillars of the future smart healthcare ecosystems that strike a balance between the clinical performance and environmental sustainability.
Bala et al. (Sun,) studied this question.