Big data is often touted as a tool for increasing quality and decreasing cost. Few people would think of it as a tool for breeding service animals. At Guiding Eyes for the Blind, a 501(c)(3) nonprofit and leading non-profit breeder and trainer of service dogs, President and CEO Thomas Panek’s vision is to harness the IBM SoftLayer cloud and Watson to increase the efficiency and flexibility of its breeding program and to produce a higher percentage of successful graduates and breed for different needs.

“We breed 500 puppies every year, of which about 150, the best of the best, are selected for guide dogs,” Panek said in a recent interview with SiliconANGLE. About 18 more dogs go to assignments serving children with autism, and others end up in arson detection or searching stadiums for unexploded ordnance with the U.S. Bureau of Alcohol, Tobacco, Firearms and Explosives.

The annual program costs about $50,000 per successful dog, Panek said. Doubling the number of successful candidates would halve that cost. That is where big data analysis and Watson come in. The group hopes that data analysis of the genetics of each dog and the traits and success it experiences over its career can increase the percentage of successful dogs going forward.

A closet full of data

The organization has huge amounts of data, including genetic records, 500,000-plus health records, 65,000-plus behavioral evaluation results, X-ray images and videos of standard tests the dogs must pass. But until Panek took over two years ago all that data was trapped in home-grown applications running on aging hardware that literally lived in closets. “Sometimes a server would crash and we would not have access for a day,” he said.

When Panek was brought in to run the organization two years ago, “the first thing I recognized walking into our IT facility was we can’t have a bird’s nest any more. IT hosting is not our core mission.” One of the first decisions he made was on a capital budget that contained a large item for replacing IT hardware. He turned it down.

Instead, he decided to leverage the cloud. He chose IBM partly for personal reasons. “After I lost my sight I was trained to program using voice recognition on an IBM 3090 mainframe, which dates me,” Panek said. “That experience showed me that IBM was way out in front on issues related to people with disabilities.”

The second reason was IBM’s long reputation for dependability. His team did consider Amazon Web Services LLC and other cloud hosting services but were nervous about trusting a startup that is not primarily a technology company “with the assets that are most precious to us, our dogs.” IBM was an easy choice.

Cleaning out the closets

The migration presented the organization with an opportunity to “clean out the closets” and normalize the data. So instead of just dumping everything into the cloud, the organization has had a team working full time for the last 18 months going through the data to make sure it is good shape as it moves to the cloud. Under Panek the organization has standardized its data collection, for instance with a standard evaluation form for assessment of the dogs that has been translated into many languages and now is being used around the world. But in the past, as with many businesses, it was not always consistent in its data collection. “There are many messy tables out there, and we’ve been working hard on cleaning them up and making sure we are migrating meaningful data.”

This is particularly important because breeding and training guide dogs is a worldwide effort. Guiding Eyes has dogs working in Hong Kong and Israel, and close partnerships with a similar breeding and training organization in Japan and with Guide Dogs UK. “When someone in Hong Kong is evaluating a dog, he uses the same form we do.”

This data is important for the breeding program. For instance, sensitivity to wearing the harness is a heritable trait, so if a dog develops sensitivity over the course of its life, that data needs to be fed back into the breeding program. Today that is done manually. In the future, Panek envisions using IBM’s analysis capabilities – and specifically Watson – to analyze the details of the records, including videos, to improve selection of the breeding lines and the success rate of future classes of puppies and decrease the cost per dog.

Specialized breeding

The requirements for dogs are also evolving. For instance, with the population aging, the organization saw a need for slower dogs to work with older people suffering from macular degeneration. Using a GPS, they determined that the dogs average between 1.8 and 3.5 miles per hour when walking. The program to breed slower-walking dogs was actually were too successful – owners became frustrated at speeds slower than 1.8 mph.

“So now we are speeding them up again,” Panek said. And doing so let to the discovery of  another niche: who are blind and love to run.” Last year, Guiding Eyes bred the world’s first official running guide dog. It’s a demanding job. Dogs have to watch out for things like tree limbs that could hit the blind runner as well as potholes and other obstacles. They need to know every step of the route and choose turns without hesitation. And they need to set a steady pace that the human can maintain distances ranging from a 10K to a marathon.

It’s a challenge that’s close to Panek’s heart. He’s a distance runner whose blonde Lab guide dog, Gus, wasn’t bred to run, but guides Panek in races up to 10K. “After about 10K he’s starting to say ‘You know, I think I’ve had my fill. Let’s go eat,’” Panek said. For longer runs, including marathons, Panek depends on a human guide. The photo above shows Panek and Gus at the end of last month’s Boston Marathon.

Word has already gotten around about the program, and the organization is getting mail from blind runners around the country who don’t have sighted partners to guide them. Panek hopes data analysis will lead to the development of other specialized lines in the future.

Having data in the IBM cloud, where it’s accessible worldwide, can lead to closer cooperation and better breeding. “If you have five people collaborating across five continents on solving one human problem, you will get a lot farther a lot faster than if you have the data on an aging server in Yorktown Heights, NY, behind a closet door,” Panek said.

Photo © Guiding Eyes for the Blind 2016