Another area that technology has the capacity to dramatically improve the status quo is in research. Researchers are using technologies such as virtual reality to study and explore areas as diverse as childhood obesity, motor skill training, and disaster response. Some of the tools developed for research may be suited for translation into health care practice; these technologies can help bridge the research-to-practice divide. In this session, three researchers—Susan Persky, Sunbin Song, and Victor Cid—spoke about their efforts to use technology in research and how these projects may translate into practice.
THE IMMERSIVE VIRTUAL ENVIRONMENT TESTING AREA
The Immersive Virtual Environment Testing Area at the National Institutes of Health (NIH), directed by Susan Persky, is a research facility with the primary goal of helping researchers integrate technology into their research programs. Persky said the program has existed for over a decade, and there is an emphasis on using technology to further research aims, not merely for the sake of technology itself. Persky told workshop participants about several advantages to using virtual reality in research.
First, virtual reality “reduces the trade-off between control and realism,” said Persky. For example, in a research study that involves a participant interacting with a provider, a virtual provider can be programmed to say exactly the same things to every patient, resulting in high internal validity. At the same time, the research can take place in a realistic clinical environment. The second benefit to virtual reality is that it allows the researcher to simulate complex or impossible scenarios; for example, a virtual provider or patient can have any characteristic—a specific race, gender, weight—that is of interest to the researcher. Persky noted that a traditional way to study these characteristics is through patient vignettes that give information about the patient’s race or gender. The virtual reality environment, where the characteristic is not called out explicitly, allows researchers to better study implicit biases. Third, virtual reality research provides an opportunity to collect specific behavioral measurements. Researchers can collect very granular data about what participants do and when they do it, which is difficult to collect in real-world environments, said Persky. Fourth, a virtual clinic experience is portable and distributable. Participants and researchers do not need to come to a specific place to conduct research; it can be done anytime and anywhere.
Persky gave several examples of how virtual reality can be used in research. One study sought to look at the relationship between a patient’s weight and how a provider treats the patient and makes decisions for her care. Researchers used two different virtual versions of “Jennifer Taylor,” who were identical in every way—including their nonverbal behavior and what they said—except for their weight (see Figure 4-1). Medical school students interacted with the patient in a controlled and realistic virtual environment, and researchers measured various parameters of interest. They found that the students made less visual contact with the obese version of the patient, and that the students believed that the obese patient was less likely to adhere to their medical advice. This suggests that “patient weight status, isolated from absolutely every other possible confound, elicits biased behavior and attitudes from physician trainees.” Persky said this “is not something that you can look at in a real-world clinical environment” (Persky and Eccleston, 2011).
Another research project that used virtual reality, said Persky, looked at how obesity risk information could influence how parents feed their children. Researchers told parents about their child’s risk for obesity, based on family history, and then asked parents to choose food for their children from a virtual buffet (see Figure 4-2). The buffet was a controlled and consistent environment—research participants always saw the same food every time, with no social pressures from other customers or differences in how fresh the food was. The virtual reality technology allowed researchers to measure which foods the parents selected, in what order they selected them, and the time delay in between their choices. Researchers validated the virtual buffet by comparing parents’ choices in the virtual world and the real world. The study found that parents who received information about their child’s specific risk of obesity put fewer calories on their food tray than the control group. In addition, parents who felt guilty about passing down a genetic risk for obesity to their child were able to reduce their guilt by choosing a healthier meal in the buffet.
Persky said that these types of research projects are being expanded to explore how manipulating the built environment (e.g., the buffet), can influence choices that people make, and to consider how other interventions or education can affect people’s actions in the virtual environment. Persky told workshop participants that one of the goals of virtual reality
research is to generate communication strategies and approaches that can be brought into practice to help benefit patients. In addition, some of these virtual reality tools may be beneficial in educational, training, or practice settings. She noted specifically that using virtual reality to detect provider biases, such as the study with the lean and obese patient, could be a good first step to a training program that would make providers aware of their bias, educate them, and then offer tools to minimize the bias.
MOTOR SKILL TRAINING USING VIRTUAL REALITY
Sunbin Song, senior researcher with the National Institute of Neurological Disorders and Stroke, told workshop participants about using interactive and immersive virtual reality for motor skill training. Song said that motor skill training is used for rehabilitation after brain injury, such as stroke or traumatic brain injury, as well as for sports training. Song noted that there has been a lot of work done with therapies that use semivirtual reality technologies, such as Nintendo Wii. However, she said that these have generally not been shown to improve patient outcomes in multicenter clinical trials because the game learning does not generalize to real-life activities (Saposnik et al., 2016). Immersive virtual reality, on the other hand, can mimic real-life tasks of daily living. Some older immersive virtual reality systems, such as CAVE (cave automatic virtual environment), are quite expensive and not easily home based; however, immersive virtual reality is quickly becoming cheaper and easier to transport.
The traditional way of studying motor learning, said Song, is through serial response time tasks (SRTTs), in which a participant clicks buttons in response to a stimulus on a computer screen. This approach has been used for more than 40 years, and has allowed researchers to characterize many aspects of motor learning, including practice-dependent and sleep-consolidation stages, sequential learning, and visual motor binding (Song and Cohen, 2014). Researchers have been able to determine neural correlates between motor learning and structural features in the brain (Song et al., 2012, 2015). Virtual reality motor skill training is based upon these findings and years of experience, said Song.
The Virtual Reality SRTT (vrSRTT), said Song, is a mixed reality experience, in which participants use a VR headset to see a virtual ball appear in front of them. The balls are differently colored, and there is a corresponding colored hole to drop the ball into (see Figure 4-3). Using vrSRTT, researchers can collect detailed data on head movement, the trajectory that the ball travels, and speed of sorting. The program can be tailored for different needs, for example, color blind-proof colors can be used for the balls. Song said that vrSRTT has been judged by participants as much more engaging than the traditional button-pressing version of SRTT. Compared to the “gold standard” of an immersive “whack a mole” game, the ball sorting program was equally natural and immersive, though less engaging. Song said that future versions of vrSRTT will be more like a game and more engaging; engagement is essential for rehabilitation because “you
want subjects to use” the game. Tests that compare vrSRTT with traditional SRTT have shown similar results in terms of participant learning.
Song reported a few preliminary conclusions that can be made about vrSRTT. First, subjects who are learning a task involving virtual objects show evidence of learning in the same way as they would with real-life objects. Second, virtual reality allows researchers to embed complex regularities into the ordering of the virtual objects and show learning in a way that would not be possible with a real-life environment. Third, the 3-D trajectory information that can be gleaned from vrSRTT is valuable for gaining insight into subtle aspects of sequential learning. The future of this technology, said Song, will involve integrating more detailed tracking of hand, head, and eye motion, developing a tennis version of the game, and creating a magnetic resonance imaging (MRI)-compatible version in order to study neural correlates. Song noted that creating an MRI-compatible virtual reality platform could be useful for diagnosing brain conditions such as traumatic brain injury, stroke, or Alzheimer’s, which currently can be difficult to diagnose through imaging. These diagnostic tools, said Song, might be developed through using algorithms or artificial intelligence to analyze big, open datasets in order to find patterns.
VIRTUAL REALITY DISASTER TRAINING
The Disaster Information Management Research Center (DIMRC) at NIH’s National Library of Medicine develops resources for disaster health preparedness, response, and recovery, said Victor Cid, senior computer scientist. One type of resource that is in development, said Cid, are simulations for training professionals in public health and disaster management. Along with collaborators, DIMRC is developing two specific projects.
The first is called Virtual Incident Command System Exercises (VIX), and it is designed to train professional staff that play roles in emergency operation centers at hospitals and other facilities. There is already an existing and widely used approach for this type of training in which participants typically gather around a table, pretend to be in a specific scenario, and role-play the various positions in the Incident Command System (ICS). However, this approach is not very engaging, said Cid, so DIMRC is leveraging technology to develop a more productive type of training. The VIX program is designed to be more engaging, realistically simulate the disaster scenario, and help participants access and use real information and communication tools as they practice their ICS roles. The use of technology will help organizations train more frequently by reducing cost and increasing the ability to have remote trainings. The VIX training simulator is like a multiplayer videogame, in which trainees gather in the simulated environment and interact with each other and the environment through
their avatars. Cid noted that DIMRC does not create and design the disaster exercise scenarios used in the simulations; rather, it collaborates with people who have already been trained to do so. VIX includes a tool to create such scenarios.
The second program in development is the Highly Infectious Disease Emergency Management (HIDEM) system. This program is designed to prepare caregivers at hospitals for an infectious disease event, and was inspired by the 2014 Ebola outbreak that resulted in several U.S. hospitals receiving patients suspected of being infected with Ebola. This program consists of a series of modules that allows caregivers to practice the main skills they will need. Modules are being developed to cover skills in donning and doffing personal protective equipment, managing waste, conducting x-ray procedures, doing lab work, and admitting procedures. The modules will allow participants to play different roles, and generally follow Centers for Disease Control and Prevention (CDC) guidelines, said Cid. For example, the doffing personal protective equipment module, which is already available, has two players, a caregiver and an observer, and the observer reads the checklist with the steps for doffing to the caregiver, who completes them with tools in the simulation. Following each step of the simulation, the participants watch a video of doffing steps and identify whether or not the step was done correctly. Afterward, the participants review any mistakes they made and receive a score.
Although these programs are still new, the preliminary results are promising. Overall, said Cid, these training programs have been well received by participants, but emergency managers tend to be “very practical” and somewhat suspicious of using computer games for training. Testing of these programs has shown higher engagement than with traditional training. Nearly 100 percent of participants would recommend adopting virtual training, and 79 percent of participants reported improvements in their preparedness attributable to the virtual training. Cid noted that using immersive virtual reality for these simulations would likely have positive benefits, but he said that barriers such as cost and user comfort need to be overcome.
While these presentations demonstrate that virtual reality is being integrated into research, Song admitted that there are several hurdles yet to be overcome to fully leverage the power of virtual reality. First, virtual reality is not being adopted quickly, owing in part to the cost as well as eye strain and simulation sickness. Second, there is a gap in knowledge; however, since virtual reality platforms can collect an enormous amount of data, knowledge about virtual reality and its application is growing rapidly.
A workshop participant asked the session speakers to comment about the research-to-practice divide, and how their work might help to bridge this divide. Persky said that some of her research projects that focus on clinical practice—such as the clinical simulator for studying weight bias—are already beginning to bridge the research-to-practice divide. In addition, she told participants about another clinical simulation that is being used to assess the efficacy of a pharmacogenetics training module. The simulation puts nurse practitioner students in a clinical situation with a patient, and examines how the student applies their classroom knowledge to patient care. Persky said these types of research will help inform education and practice, with the benefit that a virtual reality training environment allows for mistakes and learning with no harm to patients.
Song said that in her experience, technologies tend to be developed in research settings, and then are picked up by a business that develops the technology into a commercial product that can be implemented in hospitals or training centers. Song noted that researchers are not incentivized to think about the commercial or clinical application of a new technology; in fact, they are “kind of penalized.” To get grant money, researchers must focus on developing new concepts, rather than practical application of technologies. Changing this incentive structure, she said, would likely require policy change. Persky agreed that most researchers are not incentivized to think about practical application, but said that she is lucky to work in a program—the Genome Institute at NIH—where thinking about translation is a top priority. This focus allows Persky to do research that can directly translate to the clinic.
Persky, S., and C. P. Eccleston. 2011. Medical student bias and care recommendations for an obese versus non-obese virtual patient. International Journal of Obesity (London) 35(5):728–735.
Saposnik, G., L. G. Cohen, M. Mamdani, S. Pooyania, M. Ploughman, D. Cheung, J. Shaw, J. Hall, P. Nord, S. Dukelow, Y. Nilanont, F. De los Rios, L. Olmos, M. Levin, R. Teasell, A. Cohen, K. Thorpe, A. Laupacis, and M. Bayley. 2016. Efficacy and safety of non-immersive virtual reality exercising in stroke rehabilitation (EVREST): A randomised, multicentre, single-blind, controlled trial. Lancet Neurology 15(10):1019–1027.
Song, S., and L. G. Cohen. 2014. Practice and sleep form different aspects of skill. Nature Communications 5:3407.
Song, S., N. Sharma, E. R. Buch, and L. G. Cohen. 2012. White matter microstructural correlates of superior long-term skill gained implicitly under randomized practice. Cerebral Cortex 22(7):1671–1677.
Song, S., S. J. Gotts, E. Dayan, and L. G. Cohen. 2015. Practice structure improves unconscious transitional memories by increasing synchrony in a premotor network. Journal of Cognitive Neuroscience 27(8):1503–1512.