<p>IoT systems with information management in routing protocols cause significant risks to wireless sensor networks (WSN) in terms of energy distribution and productivity. Traditional routing methods lack precision and accuracy to deal with dynamic variables and data lifecycle, such as data transmission and energy consumption. Network reliability is the main challenging issues in IoT-WSN that needs to be optimized for network’s sustained operation. This study develops machine learning enabled predictive framework to assess and mitigate data loss and over consumption of energy in routing protocols specifically in enhanced cluster-based routing protocol (ECRP). Through utilization of Random Forest Regressor (RFR), and Random Forest Classifier (RFC), the model analyzes path cost estimation and cluster head selection for high predictive accuracy. To characterize IoT network activity, feature engineering has been incorporated with A star algorithm to preprocess network features explicitly, residual energy, distance metrics, and received signal strength. The study’s results indicate that network lifespan, energy consumption, packet loss proportion, and end-to-end delay contribute significantly to IoT-WSN, with predictive models focusing on energy management and multi-level data aggregation. Comparative analysis shows that neural networks outperform conventional methods, with R² value of 0.99 for regression and 0.98 for cluster head selection accuracy. The findings suggest that integration of machine learning techniques within IoT-WSN optimizes energy management, and strengthens network infrastructure.</p> Graphical Abstract <p></p>