Surgeon skill levels are the result of a combination of technical and people skills that help surgeons do their jobs more effectively. Proficiency in tissue dissection, suturing, equipment management, homeostasis, and knot tying are among the specializations. Fixation periods are examined, the gaze remains fixed, and eye tracking saccade movements in focus are employed to assess and instruct surgical skill-level categorization. The difficult aspects of such eye tracking data include congenital nerve, screen lighting, and high acquisition and maintenance costs. When provided with sufficient sample data, deep learning enables a neural network to become more proficient at a particular activity. Therefore, using eye tracking data and the surgical skill-level categorization, Robot-Assisted Surgery based on Deep Learning (RAS-DL) technologies have been improved in this study. This research looked at additional training experience levels and combined deep learning algorithms in Conventional Neural Network (CNN) with an eye-tracking approach to provide an objective instrument for evaluating surgical expertise for patients. Validated with a larger sample size and tasks, the models were used to objectively evaluate performance and learning rate in an electroencephalogram (EEG) using eye-tracking data. The finding shows the extracted features from the data to determine if characteristics from eye-tracking data were obtained while doing a peg transfer task used to identify skill levels.