TY - JOUR
T1 - Objective Assessment of Surgical Skill Using Artificial Intelligence Hand Tracking in Cardiothoracic Training
T2 - A Feasibility Study
AU - Atazadah, Morsal
AU - Bourass, Mounir
AU - Max, Samuel A.
AU - Cilon, Ivo M.
AU - Oosterhuis, J. Wolter A.
AU - Coopmans, Laurent N.A.
AU - Klautz, Robert J.M.
AU - Braun, Jerry
AU - Mahtab, Edris A.F.
N1 - © The Author(s) 2026. Published by Oxford University Press on behalf of the European Association for Cardio-Thoracic Surgery.
PY - 2026/2/1
Y1 - 2026/2/1
N2 - Objectives: Measuring surgical competency is essential for surgical residents to ensure patient safety. Traditional assessment tools rely on subjective evaluation. This study evaluated whether artificial intelligence (AI)-based hand tracking can more objectively distinguish between levels of surgical competency and predict surgical years of experience versus traditional assessments. Methods: A total of 44 participants, including medical students, surgical residents, and surgical consultants, performed transcutaneous suturing, intracutaneous suturing, and surgical knot tying. Videos of intracutaneous suturing were scored using the objective structured assessment of technical skills (OSATS). Hand movements were analysed using AI tracking software to extract coordinates to measure velocity, pathlength, and jerk. Linear regression models predicted experience years using procedural time and OSATS in combination with hand tracking metrics. Results: Hand tracking metrics varied mainly between medical students and more experienced groups. Traditional assessment tools (procedural time, OSATS) could predict experience years during training, with an adjusted coefficient of determination (R2) ranging from 0.537 to 0.638, dependent on procedure type. Hand tracking variables identified multiple significant predictors for years of experience, with an adjusted R2 of 0.540-0.712, which outperformed the traditional tools in each procedure. Combining all assessment tools (time, OSATS, and hand tracking) gave the best predictive value, with an adjusted R2 ranging from 0.540 to 0.809, with velocity, pathlength, jerk, and acceleration as significant predictors. Conclusions: AI-based hand tracking provides a new method for objective, reproducible measures of surgical skills. Incorporating hand tracking metrics enhances prediction of surgical experience, and supports standardized as well as objective evaluation of skills assessment in surgical training.
AB - Objectives: Measuring surgical competency is essential for surgical residents to ensure patient safety. Traditional assessment tools rely on subjective evaluation. This study evaluated whether artificial intelligence (AI)-based hand tracking can more objectively distinguish between levels of surgical competency and predict surgical years of experience versus traditional assessments. Methods: A total of 44 participants, including medical students, surgical residents, and surgical consultants, performed transcutaneous suturing, intracutaneous suturing, and surgical knot tying. Videos of intracutaneous suturing were scored using the objective structured assessment of technical skills (OSATS). Hand movements were analysed using AI tracking software to extract coordinates to measure velocity, pathlength, and jerk. Linear regression models predicted experience years using procedural time and OSATS in combination with hand tracking metrics. Results: Hand tracking metrics varied mainly between medical students and more experienced groups. Traditional assessment tools (procedural time, OSATS) could predict experience years during training, with an adjusted coefficient of determination (R2) ranging from 0.537 to 0.638, dependent on procedure type. Hand tracking variables identified multiple significant predictors for years of experience, with an adjusted R2 of 0.540-0.712, which outperformed the traditional tools in each procedure. Combining all assessment tools (time, OSATS, and hand tracking) gave the best predictive value, with an adjusted R2 ranging from 0.540 to 0.809, with velocity, pathlength, jerk, and acceleration as significant predictors. Conclusions: AI-based hand tracking provides a new method for objective, reproducible measures of surgical skills. Incorporating hand tracking metrics enhances prediction of surgical experience, and supports standardized as well as objective evaluation of skills assessment in surgical training.
KW - artificial intelligence
KW - cardiothoracic surgery
KW - skills assessment
KW - surgical competency
KW - surgical training
KW - technical skills
KW - Clinical Competence
KW - Cardiac Surgical Procedures/education
KW - Motion Capture
KW - Internship and Residency
KW - Intelligent Systems
KW - Artificial Intelligence
KW - Humans
KW - Education, Medical, Graduate/methods
KW - Male
KW - Feasibility Studies
KW - Female
KW - Motor Skills
KW - Hand/physiology
KW - Suture Techniques/education
UR - https://www.scopus.com/pages/publications/105031616785
UR - https://www.mendeley.com/catalogue/37894754-7e58-35f9-b772-c8fc0e95a62a/
U2 - 10.1093/icvts/ivag048
DO - 10.1093/icvts/ivag048
M3 - Article
C2 - 41669767
AN - SCOPUS:105031616785
SN - 2753-670X
VL - 41
JO - Interdisciplinary Cardiovascular and Thoracic Surgery
JF - Interdisciplinary Cardiovascular and Thoracic Surgery
IS - 2
M1 - ivag048
ER -