An IoT-enabled AI framework for sustainable product design optimizing eco-efficiency using BiLSTM.

This study aims to develop an Internet of Things (IoT)-enabled framework for sustainable product design that enhances eco-efficiency through transparent, data-driven decision-making. It addresses the limitations of conventional approaches, which often lack real-time adaptability, measurable sustainability assessment, and systematic design optimization. The framework integrates IoT sensor networks with a Bidirectional Long Short-Term Memory (BiLSTM) deep learning model to analyze real-time manufacturing data, including energy usage, material consumption, production efficiency, and environmental indicators. The BiLSTM model is benchmarked against LSTM, CNN, and traditional machine learning techniques to assess predictive performance. Robustness is ensured using five-fold cross-validation and statistical significance testing (t-test, p < 0.05). Results indicate that the proposed framework improves energy efficiency by 23.5% and reduces material waste by 19.2% compared to conventional methods. The BiLSTM model achieves a predictive accuracy of 97.6%, providing statistically significant improvements over other benchmarked models. These outcomes demonstrate reliable performance gains without overstating novelty, aligning with reviewer expectations for precise and reproducible reporting. The contribution lies in (i) applying BiLSTM-based predictive modeling to optimize eco-efficiency using real industrial IoT sensor data, and (ii) providing a transparent derivation of sustainability metrics validated on actual multi-sensor manufacturing data rather than simulated datasets. Unlike prior studies with limited real-world testing, this work evaluates the framework on real factory conditions and compares performance with established baselines. The approach is applicable across automotive, electronics, and consumer goods sectors and supports measurable progress toward the United Nations Sustainable Development Goals (SDGs).