Playing video games to stop Alzheimer’s: breakthrough or hype?

Context: This is the 4th article in a ten-week series about what I learnt at Singularity University this summer. Here are the 1st on space, the 2nd on biomimicry, the 3rd on Virtual Reality and Oculus Rift.

—

Singularity University has a natural leaning towards Biology so no wonder that, with a passion for education and learning, I would be ending up working on one of its scientific pillars : neuroscience.

The summer program is two-fold: First ‘speaking and listening’ with TED-like speakers for six weeks, then ‘doing’ through a four-week long project. Mine was called Neurofy (2:20 minute video) and targeted Dementia through video games. The idea: the latter can generate lots of new data useful to researchers and for the more optimistic of us, even limit Alzheimer’s development.

Diving into brain disorders through video games, the new gold rush?

Alzheimer’s accounts for 70% of dementia cases. It’s the only disease, among the ten deadliest in the US, that is still growing and already costs as much as 1% of the GDP in this country. It cannot be stopped, it cannot be cured, but represent a huge market ($650bn annually) and has been at the center of many technology focuses lately.

Tech-based tests are increasingly used, for Neurotrack suggested in an early test they could detect who’d have an early dementia 3 to 6 years before symptoms were visible. A growing number of other startups, such as Pzifer-backed Akili, is focusing on games to track cognitive loss of patients (and adress this gigantic market need). Cheap and actionable, those companies position those games as mass-market “medical products”: promising or hyped?

Wait. Why would health & video games meet?

A couple of gaming platforms have blossomed in the past years with a “brain fitness” positioning. Lumosity, for instance, embodies this trend and raised more than $30m lately to help its customers play little games daily and enhance their cognitive performance. Did video games become a medicine? Not too sure yet, but from gaming results to attention and body movements, they are indeed seamless data factories that help us monitor sick people or those at risk.

For example, a Spanish hospital led an experiment asking hundred of its Alzheimer’s patients to play Kinect for 9 months (video in Spanish). No miracle curing the disease but probably much more comfortable option to keep track of the patient’s condition without any painful and endless questionnaires.

Game results and behavior tracking are just a way to capture this data and should be considered as a mere interface: they have to be mashed up with other datasets to make sense. Let’s not confuse what we see on the walls of Plato’s cave and what’s happening inside the brain: even if we increasingly understand the different stages of the disease, we still have no idea why Alzheimer’s starts.

What’s next? A few learning shots to dive into neuroscience and its future steps.

A few months ago, I had no idea about most of the things I just mentioned. Here are three ideas I discovered through the super intense crash course I went through:

o Why do we learn following patterns? Simply because this is the way our brain is wired. « Learning consists of reinforcing the connections between neurons. Relational memory takes place when students learn something that adds to what they have already mastered ; they engage or expand on «maps» already present in the brain. » as explains a great book (“Research-Based Strategies to Ignite Student Learning: Insights from a Neurologist and Classroom Teacher”)

o What’s a “neural network” and why is it limited so far? Our brains are formed by connections between thousands of neurons (a “neural network”). Those connections both reinforce and die out according to what the brain uses: we call this plasticity. Computing has recently adopted this model to create machine learning capabilities: machine learning can execute a form of plasticity — i.e. the connections between the nodes (like the neurons) can be strengthened or weakened depending on the information you feed it to develop more efficiently. Nevertheless, we can’t make the “hardware” (physical chips) follow this pattern (yet?).

o Can we control a specific neuron? Even if we come have more than 100 billion of them, the answer seems to be yes. The basic principle? We can now insert a new light-sensitive protein into the calcium channels in neurons (only in mice removing their skull so far). When we shine a light on them, they “fire” and thereby we can control their activity and actions. We call this optogenetics (watch Ed Boyden’s TED talk about it for more information).

A reaction to share? Great. Reach out or comment.