Could a video game be an FDA-approvable medical device?

video game

In a recent Commonwealth Club talk, Adam Gazzaley, MD, PhD, director of a cognitive neuroscience research laboratory at the University of California, San Francisco (UCSF) described efforts to design video games that could be used for brain training, specifically in the treatment of senior mental decline as well as ADHD and autism. Researchers in the Gazzaley Lab at UCSF study the neural mechanisms of memory, attention, and perception and are evaluating innovative video games as a therapeutic approach to alleviate cognitive defects. Their "Project: EVO" platform is being productized by Akili Interactive, a Boston-based company. These video games are designed to sensitively quantitate neural function and/or intervene in any measured deficits. Akili is currently planning or conducting clinical trials using Project: EVO games in patients with ADHD (partner: Shire Pharmaceuticals), Alzheimer's disease (detection, not treatment; partner: Pfizer Pharmaceuticals), autism (partner: Autism Speaks/DELSIA), depression (partner: National Institute of Mental Health), and traumatic brain injury.

In Gazzaley's lab, researchers are evaluating a prototype, NeuroRacer, a custom-designed 3-D video game that looks like a driving simulator, in the treatment of cognitive decline in the elderly. It is designed to improve multitasking ability by training the brain to lessen the impact of interference. The idea behind this is that attention remains roughly the same at all ages, but the ability to ignore interference (interruptions and distractions, both auditory and visual) decreases with age. Thus older people tend to be poorer multitaskers than younger people. However, the brain is quite plastic and responds to training. Training in this case is delivered via a fast-paced video game that is targeted, personalized, multimodal, and closed loop. The game challenges attention, working memory, and processing speed. It is unpredictable and engaging, with adaptable difficulty, feedback (error messages, motivating rewards), richness, and immersiveness.

Gazzaley intends to seek FDA approval for NeuroRacer as a medical device. As Roland Nadler points out in his blogpost, the FDA's definition of medical device is quite broad and could encompass this technology. It would not be the first time that approval was sought for a software package: Brain Plasticity Inc, a technology incubator located in San Francisco, initiated a similar process in 2011.

A recent study by Gazzaley and colleagues, identified by the journal Nature as a "game changer", evaluated NeuroRacer in assessing and enhancing cognitive control. This game included two tasks (identifying a particular sign and keeping the car in the center of a winding road) that could be applied separately or in combination. In the assessment study, they measured the "cost" (degradation) of performance in a single task when the other task was added. When adult participants ages 20 to 85 were tested and results grouped by decade age group, a linear increase in multitasking cost was shown with each increasing decade.

The enhancement study evaluated whether NeuroRacer training could reduce multitasking cost in the 60-85 age group. Steps were taken to maintain equivalent difficulty and engagement in both tasks. For difficulty, an adaptive staircase algorithm was used to adjust the difficulty of each task independently every 3 minutes based on task performance. For engagement, rewards were given only when performance improved for both tasks beyond 80% on a given run. Older adults were randomized into 3 groups: multitasking training, single-task training, and no-contact control group; 46 subjects completed the study. Single-task training consisted of half sign-task and half driving-task training. Subjects received 12 hours of training (1 hour daily, 3 times per week for 4 weeks) followed by a post-training assessment at 1 month, and a 6-month follow-up assessment. A battery of cognitive control tests were administered, and electroencephalography (EEG) was used to evaluate the neural basis of training effects.

Results showed a statistically significant decrease in average multitasking cost from baseline to post-training only in the multitasking training group (from -64.2% to -16.2%), which did not diminish much at 6 months (still approximately -21% cost). In the single-task training and control groups at 6 months, multitasking costs had increased beyond baseline values. Similar improvements in each component skill were seen in the multitasking and single-task training groups, indicating that enhanced multitasking was not solely the result of enhanced component skills.

Significant improvements in the behavioral measures of working memory and sustained attention ("transfer effects") were reported following multitasking training that were not seen after single-task training. In the EEG analyses, two neural measures of cognitive control (midline frontal theta power and long-range coherence between frontal and posterior brain regions) were shown to be reduced in older adults at baseline compared with training-naive younger adults in their 20s. After training, significant increases (pre- to post training) were seen in midline frontal theta activity and long-range theta coherence only in the multitasking training group. Additionally, the increases in theta power seen in these individuals were significantly correlated with more sustained drop-off in multitasking cost from 1 to 6 months and with improvements in a measure of sustained attention (Test of Variables of Attention [TOVA]).

As Gazzaley points out, "The hope is that this will be the beginning of a new generation of diagnostics and therapeutics that use ... hardware and software to target neural circuits. Maybe they won't replace drugs, but drugs could be used at a much lower hopefully leading the way to think about video games for more than their role in entertainment but for their role in long-lasting and sustainable and meaningful impact."

Brain training games developed by companies like Lumosity, NeuroNation, and Rosetta Stone are a huge business, with Americans spending approximately $1.3 billion a year on them. A recent Scientific American article illustrates the doubts that scientists have about these products' promised effects. In fact, the Stanford University Center on Longevity and the Berlin Max Planck Institute for Human Development released a statement in October 2014 indicating consensus in the scientific community about these doubts. Specifically, they doubt that these brain training games "alter neural functioning in ways that improve general cognitive performance in everyday life, or prevent cognitive slowing and brain disease." At times, a study may show an interesting effect only to be shown to be irreproducible.

Even the Gazzaley study published in Nature and identified as a "game changer" on the cover, has been challenged based on its design and result analysis. Daniel Simons, professor in the Psychology Department and Beckman Institute for Advanced Science and Technology at the University of Illinois posted a detailed analysis, describing 19 shortcomings of the study. These include small sample size, possible placebo effect, transfer effects not measured at 6 months, high dropout rate (23%) due to not being able to perform study tasks, no correction for multiple comparisons in the transfer measure analyses and results somewhat inconsistent in relation to the conclusion, no means or standard deviations (SD) included, study blinding not stated, and possible sources of study bias not addressed (eg, differences in computer experience among subjects). Simons concludes that although the study is worthwhile and is not fundamentally flawed, more work is needed to confirm a therapeutic benefit.

Blogger: Ginny Fleming, Founder, Lucidize Medical & Scientific Editing. Chief capacities: medical, scientific, and technical writing and editing.

Image courtesy of Idea go at