The Power of Play
From Wii Fit to Dance Dance Revolution, the gamification of physical therapy is becoming increasingly popular, and with positive results.
A controlled clinical trial in 2013 investigated the feasibility, acceptability, and safety of a supervised video game exercise program administered via Dance Dance Revolution in individuals with Huntington’s disease.
Through a series of outcome measures that included spatiotemporal gait measures, four-square step test and activities-specific balance confidence scale, findings showed that most participants improved on game play, enjoyed playing the game, and wanted to continue playing after study completion. After playing Dance Dance Revolution, participants showed significant reductions in double support percentage (adjusted mean difference (95% confidence intervals): -2.54% (-4.75, -0.34) for forward walking and -4.18 (-6.89, -0.48) for backward walking) and those with less severe motor symptoms had reductions in heel-to-heel base of support during forward walking.
One of my areas of focus at Cionic is the creation of movement controlled games. Our efforts began organically as a showcase for on-device Machine Learned models that can predict human movements from our wearable sensor data. With the ultimate goal of augmenting human movement, we need to first predict movement intent. Games provide a great opportunity to collect well labeled movement data, and an environment to test the accuracy of our real time models.
Dance Dance Revolution is a perfect candidate. Forward, side, and back steps translate practically to training exercises for balance and mobility. In order to make predictions, we needed a model, and in order to build a model, we needed data…a lot of data. I strapped myself up with Cionic sensors and began stepping forward, to the side, and back in response to a digitally prompted cadence. I also wore foot pressure sensors, which were used to label the time bounds of each step. More than a thousand steps later, and with many training iterations, we had models that would predict step direction based on electromyography (signals of muscle firing) and kinematics (how my body moves through space). The initial models performed well, and each optimization made them nearly perfect…at least on simulations of data previously collected.
Now for the fun part! To test the performance of our models, we loaded them onto the Cionic device and used them to control an open source version of DDR software called “Endless Revolution.” Basically, the model took my body signals as inputs and spit out step directions as outputs (see visual below). Adding some boogie-worthy tracks was the final touch before I was ready to play!
Seeing those arrows light up when I stepped out to the floor was magical…like stepping on an invisible button. It was the first time I truly experienced my body being digitally augmented. There was no game controller to interact with — I was the controller. Despite the game’s simplicity, I lost myself in wonder thinking about how an algorithm could understand where my foot was going to go based on signals sent from my brain, down my spine and through my nerves, and intercepted by our device as they transmit to the muscles in my legs. It’s still a little mind blowing to think about.
The magic and wonder of Endless Revolution inspired the team to create an SDK that could enable any game to be controlled by the body.
We made a few guiding principles along the way:
- All games need to accommodate a wide range of movement abilities so that they are fun and functional no matter where someone is on their mobility journey. For someone with little direct experience working with people with ability differences, it’s easy to forget just how difficult some movements can be. I learned the importance of observing other people with different levels of ability. We built completely customizable triggers to adapt to anyone. Games can be played by flexing an ankle, contracting a calf muscle, leaning forward, and limitless other inputs.
- Game setup needs to be quick and easy or the barriers to playing become a frustrating turn off — like the arcade that keeps spitting out your quarter unless you insert it just right. You’ll give up and play something else if it’s too hard to play. We streamlined the UI to make setup more intuitive. We also made controller profiles for each user to reduce setup time and get people into games faster.
- Maximize extensibility of the game platform. Our vision to help overcome disability is bigger than any one company can accomplish. That is why we are building open APIs to enable the community to build more games on the platform. The more accessible games that can be published to Cionic’s ecosystem, the better we can superpower the body.
Today we are playing pinball with our feet as flippers, navigating a maze with the orientation of our leg, and skiing down the mountain by shifting our weight. Most importantly, the games are fun for everyone regardless of skill or ability. More than once I have become so immersed in play that I forgot it was even exercise, until I woke up the next morning, sore in muscles I didn’t even know I had. This is the power of play, and why we strive to build engaging experiences to make therapy fun.
What began as a feasibility exploration has now become an integral part of our platform. These first few games are an exciting milestone for us, but they represent just the tip of the iceberg as far as the kinds of engaging experiences Cionic can unlock. I’m excited to expand our gaming capabilities to unlock more thrilling and immersive games that help people transcend physical limitations.
If you’re a developer who’s interested in building accessible games on the Cionic Platform, I’d love to hear from you: dean@cionic.com
