Use of a machine learning algorithm to classify expertise: analysis of hand motion patterns during a simulated...

Area of Science:

• Biomedical Engineering
• Machine Learning in Surgery
• Surgical Skill Assessment

Background:

• Objective assessment of surgical expertise is crucial for training and patient safety.
• Hand motion patterns during surgical tasks contain valuable information about skill level.
• Traditional methods for assessing surgical proficiency can be subjective and time-consuming.

Purpose of the Study:

• To evaluate the hypothesis that machine learning algorithms enhance the prediction of surgical expertise.
• To classify surgical expertise based on surgeons' hand motion patterns using machine learning.
• To compare the predictive power of a Support Vector Machine (SVM) algorithm against a Lempel-Ziv (LZ) complexity metric.

Main Methods:

• Hand motion patterns were captured using inertial measurement units during simulated venous anastomosis tasks.
• Patterns were preprocessed into symbolic time series and labeled as expert (attending) or novice (resident).
• A Support Vector Machine (SVM) algorithm and Lempel-Ziv (LZ) complexity metric were used for classification.

Main Results:

• The Lempel-Ziv (LZ) complexity metric achieved 70% accuracy in classifying hand motion patterns.
• The Support Vector Machine (SVM) algorithm demonstrated higher accuracy at 83% for classification.
• SVM showed improved sensitivity (86%) and specificity (80%) compared to LZ metric for expert/novice discrimination.

Conclusions:

• The Support Vector Machine (SVM) algorithm significantly increased the predictive power for classifying surgical expertise from hand motion.
• The study confirms that machine learning can objectively differentiate between expert and novice surgeons based on motion patterns.
• This system holds potential as a cost-effective tool for objective procedural proficiency assessment in surgical training.