Title: Automated Coaching using Surgical Data Science: Can Crowdsourcing Help?
Speaker: (Anand Malpani, Johns Hopkins U.)
Date/Start Time: Thursday 5/3, 1:15pm - 2:30pm
Location: UTA 1.204
Surgical education community advocates for lifelong learning, coaching, and competency-based rather than experience-based assessment and certification of surgeons. However, current approaches to achieve these goals require time consuming review of performance by faculty surgeons, which is not scalable. Automated methods using machine learning techniques analyzing video images, instrument and hand motion, and sensor data, can provide efficient and immediate assessments on a routine basis enabling individualized feedback and coaching.
In this presentation, I will focus on whether crowdsourcing can be used to gather information about surgical skill in a given performance to drive such automated coaching. I will present results on multiple datasets from surgical training tasks across multiple surgical specialties wherein we show that crowdsourcing generates reliable and valid skill assessments. I will compare two survey types for crowdsourcing surgical skill - absolute ratings and relative ratings. I will present results on machine learning methods trained to automatically predict such crowdsourced surgical skill information. Finally, I will describe a simple adaptive approach to crowdsource surgical skill ratings to reduce costs associated with worker payments in comparison to brute force crowdsourcing.
Anand Malpani is an assistant research scientist at the Malone Center for Engineering in Healthcare at the Johns Hopkins University. His research focus is on development of technology platforms to augment human learning of skilled activities. His current work is in the realm of surgical education and training, where he is developing an automated virtual coach. He uses crowdsourcing and machine learning methods along with virtual reality simulation in this virtual coaching framework to identify deficits in performance and deliver directed feedback to the trainee. He collaborates with engineering faculty, clinicians at the medical school, cognitive scientists in school of arts and sciences, and faculty in the school of education, as well as industry partners at Intuitive Surgical Inc. and SenseGraphics AB. He is a key personnel on NSF and NIH awards and co-investigates grants awarded through the Johns Hopkins Science of Learning Institute and the American Hernia Society Foundation.
In 2017, he completed his PhD in computer science at the Johns Hopkins University under the advising of Prof. Gregory Hager. His thesis developed methods for an automated virtual coach for surgical skills training. Prior to this, he graduated with a B.Tech. in Electrical Engineering from the Indian Institute of Technology Bombay in 2010. He was awarded the Intuitive Surgical Student Fellowship in 2013, and the Link Foundation's Modeling, Training and Simulation Fellowship in 2015. He was a summer research intern in the simulation team developing the da Vinci Skills Simulator, a training platform for robot-assisted surgery, at Intuitive Surgical Inc. (Sunnyvale, CA) in 2015.
1:15pm to 2:30pm