Automated assessment of non-technical skills by heart-rate data

Area of Science:

• Medical Education
• Surgical Skill Assessment
• Physiological Monitoring

Background:

• Current surgical skill assessment methods, including observer-based and automatic systems, have limitations.
• Observer-based assessments are subjective, while automatic methods often focus on technical skills.
• There's a need for objective assessment of non-technical surgical skills using less direct data.

Purpose of the Study:

• To explore the use of heart-rate data for predicting observer-based surgical skill scores.
• To investigate the potential of random forest regressors in analyzing physiological data for skill assessment.
• To assess non-technical skills using data not directly related to technical performance.

Main Methods:

• Collected heart-rate data from 35 junior orthopedic surgical residents performing a meniscectomy on a simulator.
• Utilized the Arthroscopic Surgical Skill Evaluation Tool (ASSET) score, assessed by two independent observers.
• Preprocessed heart-rate data and extracted 41 features, then optimized a random forest regressor for score prediction.

Main Results:

• Achieved promising predictions for partially non-technical skill components, notably safety (mean absolute error of 5.76%).
• Identified key features correlating with ASSET components, suggesting links to sympathetic and parasympathetic nervous system activity.
• Demonstrated the potential of heart-rate variability analysis in surgical skill assessment.

Conclusions:

• Heart-rate data, analyzed with random forest regressors, can aid in automatic surgical skill assessment, particularly for non-technical aspects.
• This physiological data approach offers a novel, objective method for evaluating surgical proficiency.
• Combining heart-rate data with kinematic data may lead to more accurate and comprehensive automatic skill assessment systems.