Automated robot-assisted surgical skill evaluation: Predictive analytics approach.

Area of Science:

• Robotics in Surgery
• Surgical Skill Assessment
• Machine Learning Applications

Background:

• Traditional surgical skill assessment relies heavily on subjective evaluations.
• Advancements in robot-assisted surgery offer potential for objective performance metrics.
• This study introduces a predictive framework for objective skill evaluation using movement data.

Purpose of the Study:

• To develop a machine learning classification framework for automatic surgical skill evaluation.
• To differentiate between surgeons of varying expertise levels (novice vs. expert).

Main Methods:

• Extraction of eight global movement features from da Vinci robot trajectory data.
• Application of three machine learning classification algorithms: k-nearest neighbors, logistic regression, and support vector machines.
• Training and testing the framework on data from novice and expert surgeons.

Main Results:

• The proposed framework achieved high accuracy in classifying surgeon expertise.
• Accuracy rates of 82.3% for knot tying and 89.9% for suturing tasks were recorded.
• Demonstrated the effectiveness of movement trajectory data for skill differentiation.

Conclusions:

• Machine learning methods can effectively automate the classification of surgical expertise.
• Global movement features derived from robot-assisted surgery are valuable indicators of skill level.
• This framework provides a foundation for objective and reliable surgical performance assessment.