Sensors, motor tracking and data science: The quest to train MDs like elite athletes

Science Highlights
April 11, 2017
Raymond A. MacDougall

WALS speaker builds simulators to give feedback in medical training

Elite athletes perfect their A game with electronic feedback they and their coaches may analyze of swim strokes, golf swings, and numerous other motions and behaviors. Dr. Carla Pugh, guest speaker at the March 8, 2017, NIH Director’s Wednesday Afternoon Lecture Series, recognized early in her medical career that like athletes, physicians could benefit from the data obtained with sensors and motor tracking devices to learn and improve their technique. Pugh is the Susan Behrens Professor of Surgical Education and vice-chair of education and patient safety at the University of Wisconsin-Madison.

Pictured following the Wednesday Afternoon Lecture are, from left, guest speaker Carla Pugh; Roderic I. Pettigrew, NIBIB director; and Grace Peng, director of the NIBIB program in computational modeling, simulation and analysis.
Pictured following the Wednesday Afternoon Lecture are, from left, guest speaker Carla Pugh; Roderic I. Pettigrew, NIBIB director; and Grace Peng, director of the NIBIB program in computational modeling, simulation and analysis. 

As a high school athlete, Pugh recalls feedback she received after every match. By contrast, following medical school and a surgical residency, she found that the concept of feedback got left behind. She returned to the classroom for a Ph.D. in education, including coursework in human-computer interfaces, motivated by the burning question: “How can we get the data such that no surgeon in training or in practice would ever practice their craft without having continuous feedback on how to be the best that they can be?”

In 17 years of research, the past 10 as an NIH grantee, Pugh has designed simulators with sensors and detectors that provide an objective assessment—through data—of medical skills. Clinicians in training and seasoned experts alike have lined up to user her stimulators, contributing valuable performance data and gaining valuable feedback with their participation.

Pugh provided an overview of the evolution of her use of simulators in medical training. Her first, in 2000, was built with medical mannequins outfitted with rudimentary sensors to record touch and pressure delivered in a simulated gynecologic exam. Medical students performing simulated exams appeared competent, but the sensors and computer interface revealed quite different touch and pressure readings that guided further instruction for each student. “That’s when I knew there was something huge here—something we could potentially do with this data,” Pugh said.

Force sensitive resistor
When placed onto a manikin or surface, flexible sensors can detect touch and degree of pressure applied for medical exam simulations. 

Pugh expanded that simulator study to include 700 expert clinicians in the obstetrics and gynecology field. By comparison, experts completed the simulated exam in half the time and with less pressure and greater accuracy. That predictable result was accompanied by the a surprise effect of experts lining up to be a participant so they could receive the feedback provided by the simulation.

The laboratory’s lineup of simulators soon included a digital rectal exam, which proved to be a valuable medical training assist. “It was quite interesting to see the students looking at the computer while learning the exam, and autocorrecting without needing feedback from a faculty member.” Pugh said. “One of the things that we discovered is that there is a lot lost in translation when the students go from reading what they’re supposed to do in a text book and actually trying to do it.” Her team received an award for the work in 2008 from the Association for Surgical Education. 

“I began to instrument any simulator I could get my hands on,” Pugh said in describing the intubation simulator. The tricky procedure is performed for surgeries as well as to establish an airway for a patient during an emergency. “We put sensors in places where you should be touching and where the instruments should touch, and we put them in places where you shouldn’t.” The data showed that an experienced doctor could place an intubation tube in 12 seconds with fewer wrong moves than a novice committed with 18 added seconds.

Breast exam simulation followed. “It’s actually kind of complex; everyone does it differently,” Pugh said. In 2011, the United States government recognized Pugh’s breast exam project, supported by the National Institute of Biomedical Imaging and Bioengineering (NIBIB), with a Presidential Early Career Award for Scientists and Engineers (PECASE).

The breast exam simulator represents a significant advance from Pugh’s early simulators. Sensors in the device can detect directional force and her team videotaped each simulation to be synched with force and directional data. This simulation study showed that 15 percent of physicians who participated with an average of 20 years in practice do not apply enough force to find a lesion in a breast exam. More than half didn’t practice the recommended linear strip palpation that achieves greatest tissue coverage.

Pugh’s ongoing projects extend to operating room procedures, from sutures to laparoscopy. She collaborates with experts in to differentiate data signals for various types of touch, and with data scientists to extract the many performance variables from her data.

“There is an endless supply of sensor data and things that we can try in the operating room,” Pugh said. “The simulator is my bench, my mouse. I can have different simulators, but it’s the data that is helping move this agenda and the science forward.” 

Explore More

There is so much more to explore at NIBIB.