The foundations of Wireless Sensor Networks (WSNs) are diminished when they face unexpected node failures, and because of limitations on the energy available at individual nodes. In this study, an intelligent routing framework is presented that integrates machine learning techniques with the concepts of energy-awareness to provide improved routing in the WSN environment. The Adaptive Routing Algorithm proposed here is based on multiple routing performance metrics, including latency, residual energy, packet delivery success rate, and probability of node failure. These performance metrics are obtained through the use of a Random Forest classifier. As a result, by avoiding routing through potentially faulty nodes, this routing algorithm improves the reliability of the routing structure and provides a better balance of energy consumption and packet loss. The results of the simulation conducted demonstrate that using the proposed routing algorithm can achieve up to 98% packet delivery ratio, approximately 8.5ms end-to-end latency, and extend WSN network lifetime by as much as 8% when compared to protocols based on Min Transmission Energy and traditional Adaptive Routing algorithms. Therefore, the results indicate that combining machine learning with energy-aware routing improves QoS and extends the lifetime of Wireless Sensor Networks.