TAIPEI, TAIWAN (UroToday.com) - Introduction and Objectives: Crowdsourcing is the practice of obtaining services from a large group of people; typically from an online community such as the Amazon.com Mechanical Turk Project. We hypothesized that the ‘crowd’ could score performances comparably to scores derived from expert surgeons of dry lab laparoscopic skill tasks videotaped during the AUA BLUS curriculum validation project.
FREE DAILY AND WEEKLY NEWSLETTERS OFFERED BY CONTENT OF INTEREST
Did you find this article relevant? Subscribe to UroToday-GUOncToday!
The fields of GU Oncology and Urology are advancing rapidly including new treatments, enrolling clinical trials, screening and surveillance recommendations along with updated guidelines. Join us as one of our subscribers who rely on UroToday as their must-read source for the latest news and data on drugs. Sign up today for blogs, video conversations, conference highlights and abstracts from peer-review publications by disease and condition delivered to your inbox and read on the go.
Methods: 24 candidate videos of laparoscopic skill tasks performed by surgeons of varying levels of laparoscopic case experience - 12 suturing and 12 pegboard transfer performances were evaluated by 5 faculty experts and at least 60 Amazon.com Mechanical Turk crowd-workers. Each rater provided responses to the same multi-domained rating scale from the Global Objective Assessment of Laparoscopic Skills (GOALS) tool. We compared mean global performance scores provided by experts and crowd-workers using Cronbach’s alpha and estimated performance-specific passing probabilities by cut-offs established with receiver operating characteristic (ROC) curves.
Results: Within 48 hours we received 1,840 crowd-worker ratings, of which 1,438(78.2%) passed analysis eligibility criteria based on discrimination questions used to assess the integrity of the scorer’s responses. Faculty experts completed the reviews in 10 days. C-SATS ratings provided excellent discrimination between passing and failing video performances as defined by faculty experts (area under ROC curve = 96.9%; 95% CI: 90.3%–100%).
Conclusions: A properly-sized and qualified crowd can accurately score laparoscopic skill performances on par with faculty experts. Crowd-based ratings may be an efficient method for assessing passing/failing performances, and for measuring change in performance after training.
Source of Funding: None
|Listen to an interview with Thomas Lendvay, one of the authors of this study.|
Presented by Thomas Lendvay,1 Bryan Comstock,1 Timothy Averch,2 Geoffrey Box Bodo Knudsen,3 Timothy Brand,4 Michael Fernandino,5 Jihad Kaouk,6 Jaime Landman,7 Benjamin Lee,9 Elspeth McDougall,9 Ashleigh Menhadji,8 Bradley Schwartz,10 Robert Sweet Timothy Kowalewski11 at the 32nd World Congress of Endourology & SWL - September 3 - 7, 2014 - Taipei, Taiwan
1Unviersity of Washington, USA
2University of Pittsburgh, USA
3Ohio State University, USA
4Madigan Army Medical Center, USA
5Duke University, USA
6Cleveland Clinic Foundation, USA
7University of California, Irvine, USA
8Tulane University, USA
9University of British Columbia, Canada
10Southern Illinois University, USA
11University of Minnesota, USA