Whether or not researchers believe that brain training with computer games improves cognition, some have discovered a way to mine this vein for clinical research. Lumosity, the most prominent of the brain-training companies, has opened its gaming database to researchers as part of what it calls the Human Cognition Project. The database contains records of every mouse click from the 1.5 billion game sessions logged by the company’s 60 million users. Researchers from around the globe are sifting through it in search of patterns that may illuminate how aging, stress, sleep, and other lifestyle factors affect cognition. In particular, some scientists are panning the database for signatures of early decline that may flag people for prevention trials. Some researchers think brain games in general—which adapt to each user’s cognitive ability—may one day serve as cognitive diagnostics to monitor progression or help enroll clinical trials. Whether brain training boosts brain performance remains unclear (see Part 1 of the series), but in the meantime researchers are moving ahead to put the game data to some serious use.
Michael Weiner of the University of California, San Francisco, hopes to use the game data to identify people who show signs of cognitive decline and may make good candidates for AD prevention trials. Weiner will present initial results at the Alzheimer’s Association International Conference in Copenhagen, Denmark, in July. He told Alzforum that preliminary findings suggest that game performance slips with age, and that this is more severe in some people than others. Alas, navigating the data was no small task, Weiner said. “The data has a lot of variability. People are gaming in an unsupervised setting, and there are many factors that affect game performance,” he said. While Weiner hopes the game data will be useful, “we are just at the beginning of the beginning,” he said.
The Lumosity database offers the advantage of a historical record of cognitive function, noted Weiner. “When you enroll people in clinical trials, you have very little history about them,” he said. “Having some kind of objective, longitudinal data, even if it’s not perfect, can potentially identify people who are at risk for future cognitive decline and AD.” As therapeutic trials targeting preclinical Alzheimer’s are gearing up, they are facing the challenge of identifying people who are sliding toward Alzheimer’s dementia but are not showing up in doctors’ offices yet. At present these trials use various combinations of genetic and fluid and imaging biomarkers to find the right trial participants. If valid cognitive performance predictors can be found in the Lumosity database, then in theory the portion of its membership older than 50 might provide the necessary cohorts for some of those trials. Lumosity declined to share how many of its users are over 50, except to state that its user demographic is “similar to the demographic of the Internet.” According to the Pew Research Internet Project, 88 percent of people over 50 and 57 percent of people 65 and older used the Internet in 2014.
Murali Doraiswamy of Duke University in Durham, North Carolina, was one of the first researchers to partner with Lumosity. His work took a different tack. In a study that incorporated the cognitive performance of more than 100,000 Lumosity gamers per experimental group, Doraiswamy and Lumosity researchers reported that people who get seven hours of sleep per night perform optimally on the games, and that people who indulge in one to two alcoholic drinks per day outperform those who have more drinks or none (see Sternberg et al., 2013). These findings fall in line with previous studies that used cognitive tests, rather than games, to measure similar factors. Research on sleep and cognitive decline in aging reported seven hours of sleep to be ideal (Devore et al., 2014), and the finding that one or two alcoholic drinks, but not more or none, are relatively protective, is confirmed by meta-analysis (see AlzRisk).
While some researchers argue that the data from cognitive tests taken in uncontrolled settings is contaminated by uncertainties, Doraiswamy considers the home setting to be closer to reality than what researchers see in the clinic. “The Holy Grail in assessing cognitive problems, or assessing whether a person is showing early signs of AD, is to judge them in the settings where they are having problems,” Doraiswamy said. “That’s the real world.” On that, Weiner agrees. “We will present data at AAIC supporting the view that at home unsupervised tests are pretty good,” he said.
Even so, Weiner sees the Internet as a cost-efficient recruitment aid for clinical trials. He recently kicked off the Brain Health Registry, a UCSF-run website that collects demographic and longitudinal cognitive data on people of all ages who sign up to join in the research effort. The registrants answer questions about their mental and physical health, then take cognitive tests every six months. Rather than using games to assess cognitive function, this registry employs the CogState Brief Battery, which measures working memory, attention, and processing speed. The Brain Health Registry also uses the Brain Performance test from Lumosity, which advertises this registry on its own website.
Weiner plans to present the initial findings from the registry in Copenhagen as well, but told Alzforum he is elated by the speed of enrollment and how well the online cognitive test data seems to match up with similar data collected in a clinical setting. “We’ve had an unbelievably phenomenal success. Seven thousand people enrolled in three months, and more than half took the CogState battery.” Academic institutions and pharmaceutical companies can tap the registry for potential study enrollees, Weiner said. He added that the online tests do not replace conventional testing done by neuropsychologists. Instead they can serve as an efficient screen and offer an unprecedented level of longitudinal monitoring. “The question is, how can the Internet help us get to a cure more rapidly?” Weiner asked. “I think there are many ways to use it, and this is one.”
Putting Games to the Test
Apart from their potential roles in boosting cognition or flagging people in cognitive decline, some researchers propose that brain-training games may also serve as sensitive cognitive tests in their own right. Cognitive tests used in AD clinical trials, such as the MMSE, CogState, and batteries such as the ADAS-Cog, are highly validated and administered under the supervision of a neuropsychologist. Advances are being made in putting together computerized composites, which researchers hope will be sensitive enough to detect people in earlier stages of cognitive decline (see Jun 2014 news story, Nov 2013 news story, and Alzforum Webinar).
However, unlike games, current tests used in clinical trials are not adaptive, meaning they do not adjust their difficulty in real time based on how well a person performs. This leads to varying amounts of effort and frustration on the part of people taking the test. It also leads to “ceiling” and “floor” effects, where the tests are either too easy or too hard to detect change in some people. “There are many tools that have been well established in AD, but a lot of them are too easy for people with MCI,” said Veronika Logovinsky of Eisai, Inc. in Woodcliff Lake, New Jersey. However, while Logovinsky thinks adaptive tests are likely to become the norm someday, she said implementing an adaptive test or game as a cognitive measure in a trial is an uphill battle. “Trials are a very tricky business, so if you introduce something new, you have to convince regulators that it provides valid results,” she said. “Right now, it’s clearly easier to just give the same test for everybody.”
Despite the potential for regulatory hurdles, Adam Gazzaley of the University of California, San Francisco, thinks cognitive tests based on adaptive games may one day prove themselves exquisitely sensitive. “We call it a game, but when you distill it down to its active ingredients, it’s a carefully calibrated cognitive test, and the adaptivity element helps to make it that way,” Gazzaley said.
Keith Wesnes of Wesnes Cognition Ltd. in Streatley on Thames, England, is an expert on the use of cognitive testing in dementia trials (e.g., Wesnes et al., 2013). He agreed that carefully designed computerized games could hold potential as tests. “It’s not far-fetched to believe that you could have a game that measures your cognitive function,” Wesnes said. “The game could be used as a way to measure someone’s cognitive changes over time.”
At least one company is hoping the games will win the hearts and minds of regulators and pharmaceutical companies seeking to detect subtle changes in cognition in response to drugs, or to pick out participants for such trials. Gazzaley co-founded the Boston-based company Akili Interactive Labs. This startup is developing a revved-up version of Gazzaley’s NeuroRacer game, called Project Evo, for cognitive training as well as diagnostic purposes. In collaboration with Pfizer, Akili plans to enroll 100 people who have subjective memory complaints that fall short of cognitive impairment measurable by the Mini Mental State Exam or ADAS-Cog. The participants will play the game regularly at home over the course of a month, while researchers monitor their every move from afar. The goal of the study is to determine whether a participant’s low performance and learning of the game correlates with positive AD biomarkers such as PET scans for amyloid, MRI for cortical thinning, and blood and genetic biomarkers. “If it works, then you have a cheap, at-home segmentation tool to refer people to clinical trials or to start taking early interventions,” Eddie Martucci of Akili told Alzforum at the NeuroGaming conference. The study is set to begin within the coming months, Martucci said.—Jessica Shugart
- Is Brain Training More Than Just Fun and Games?
- Test Battery Picks Up Cognitive Decline in Normal Populations
- Are New Cognitive Tests Ready For Preclinical Trials?
- Sternberg DA, Ballard K, Hardy JL, Katz B, Doraiswamy PM, Scanlon M. The largest human cognitive performance dataset reveals insights into the effects of lifestyle factors and aging. Front Hum Neurosci. 2013;7:292. Epub 2013 Jun 20 PubMed.
- Devore EE, Grodstein F, Duffy JF, Stampfer MJ, Czeisler CA, Schernhammer ES. Sleep duration in midlife and later life in relation to cognition. J Am Geriatr Soc. 2014 Jun;62(6):1073-81. Epub 2014 May 1 PubMed.
- Wesnes KA, Edgar CJ. The role of human cognitive neuroscience in drug discovery for the dementias. Curr Opin Pharmacol. 2014 Feb;14:62-73. Epub 2013 Dec 22 PubMed.
No Available Further Reading