Systematic approach for content and construct validation: Case studies for arthroscopy and laparoscopy
Autor: | Venkata Sreekanth Arikatla, Sinan Kockara, Jake Farmer, Shahryar Ahmadi, Kevin W. Sexton, Tansel Halic, Recep Erol, Daniel Ahmadi, Doga Demirel |
---|---|
Rok vydání: | 2020 |
Předmět: |
Computer science
Biophysics Video feedback Virtual reality Machine learning computer.software_genre Article Feedback Arthroscopy User-Computer Interface 03 medical and health sciences 0302 clinical medicine medicine Humans Computer Simulation Laparoscopy medicine.diagnostic_test business.industry Rank (computer programming) Construct validity Computer Science Applications Support vector machine Statistical classification 030220 oncology & carcinogenesis 030211 gastroenterology & hepatology Surgery Clinical Competence Artificial intelligence business computer |
Zdroj: | Int J Med Robot |
ISSN: | 1478-596X 1478-5951 |
Popis: | Background In minimally invasive surgery, there are several challenges for training novice surgeons, such as limited field-of-view and unintuitive hand-eye coordination due to performing the operation according to video feedback. Virtual reality (VR) surgical simulators are a novel, risk-free, and cost-effective way to train and assess surgeons. Methods We developed VR-based simulations to accurately assess and quantify performance of two VR simulations: gentleness simulation for laparoscopy and rotator cuff repair for arthroscopy. We performed content and construct validity studies for the simulators. In our analysis, we systematically rank surgeons using data mining classification techniques. Results Using classification algorithms such as K-Nearest Neighbors, Support Vector Machines, and Logistic Regression we have achieved near 100% accuracy rate in identifying novices, and up to an 83% accuracy rate identifying experts. Sensitivity and specificity were up to 1.0 and 0.9, respectively. Conclusion Developed methodology to measure and differentiate the highly ranked surgeons and less-skilled surgeons. |
Databáze: | OpenAIRE |
Externí odkaz: |