Conventionally, monitoring the depth of anesthesia (DoA) is performed using the standard procedures of either observing the patient’s vitals or through commercial electroencephalogram (EEG)-based monitors. The reports of intraoperative awareness indicate that these methods are still unreliable for all patients and anesthetic agents. This paper proposes a novel approach for accurate DoA estimation based on Stationary Wavelet Transform (SWT) and fractal features while utilizing Multilayer Perceptron (MLP) Classifier, a class of Feed Forward Neural Network. The classifier utilizes an optimized temporal, fractal and spectral feature set to identify the patient conscious level irrespective of age and the type of anesthetic agent. The proposed algorithm is validated on 95 patients (age 5 months-67 years), (weight: 6 - 90 Kg). The anesthetic agents used in this study include Propofol, Sevoflurane, Isoflurane, Fentanyl, Ketamine, and Caudal. The proposed DoA classifier outperforms the state-of-the-art DoA prediction algorithms with the highest accuracy of 96.8% while utilizing minimized feature set and a deep learning-based MLP classifier for the first time in literature.