Cognitive that cares: from Alder Hey to CogniToys, here’s how AI is helping kids in the real world

“Humans only have so much bandwidth, and as patient complexity increases you either have to add more doctors – which we don’t have – or do a worse job, neither of which are options,” says Iain Hennessey, clinical director of innovation at Alder Hey Children’s Hospital. “If we can get artificial intelligence to address the bottom 20 per cent of the thought automation-type tasks, we can deal with this rising complexity. Otherwise we’re stuffed.”

Thanks to a joint £315m initiative into big data applications between IBM and the Science and Technology Facilities Council (STFC), Alder Hey is working closely with Cheshire’s STFC Hartree Centre to apply IBM’s Watson to improve the patient experience. “It is very difficult to treat a child who is scared or unhappy,” says Hennessey.

The first phase of the five-year collaboration launched in May 2016, accelerated by a fortuitous meeting at The Drum’s Do It Day – the 24-hour event in which marketers collaborate to solve real-world challenges. The event saw the Watson team connect with Jungle, an agency that had been creating content for Alder Hey. As Phil Westcott, the European ecosystem leader at IBM Watson Group, says: “The opportune meeting allowed a more joined-up way of working with Alder Hey.”

Each patient’s appointment letter now includes a QR code that, when scanned, downloads the Alder Hey app. This is used to gather voluntary information from patients and their families relating to their expectations and experiences before, during and after being in hospital. The app answers any questions that parents and children have about their stay, and gathers information such as the child’s favourite colour or TV programmes in order to build a patient profile. Watson uses natural language processing and machine learning to gather insights from such large data sets.

In the quest to continually improve the patient experience, Hennessey cites the example of carrying out heart scans, which are essential for planning whether a child needs surgery, or informing which drugs they require. “To carry out the scan they need to sit still and relax for half an hour while we smear their chest with jelly and use big probes. You can imagine how difficult that can be with a three-year-old.”

And so, a short-throw projector now screens images onto each wall of the echocardiogram room, tailored to the preferences of each child. “It can turn it into an area for under threes or a fairground, or project the child’s favourite character, for example,” says Hennessey. “The kids love it – they jump on the bed and lie totally still while we do the scan.” The benefits are manifold, according to Hennessey. “You have happy kids and happy parents, but you also get the scan done in half the time, and you get a better quality scan, which means the patient gets the right treatment.”

Watson will eventually be able to automatically tailor the nature and length of the content to each child and their procedure, and Hennessey is currently talking to content providers with a view to growing its library of material.

Watson is also able to understand not just the content of written text but also its intent, sensing anxiety by analysing word choice and sentence structure. “It’s fascinating. It’s a computer but it does it better than most humans because it can dispassionately look at language and the way it has been written,” says Hennessey. This is enabling the hospital to pick up undue anxiety and stress before a patient arrives, allowing it to intervene at an earlier stage.

It is early days, with Hennessey saying Alder Hey is some 12 months away from starting to gauge the true impact of Watson, but that doesn’t stop him from thinking big, and he has a clear vision of 20 years down the line.

“I want to give the hospital almost lifelike attributes, with a sensory system, brain and caring heart, so the building itself can holistically look after the kids within it.”

A child could walk into the hospital and the building would ‘sense’, for example, that their temperature is up and their respiratory rate is elevated, intuitively knowing which category of triage they require.

“It would simply automate the low-level intelligence part, and it’s a model that could be replicated around the world,” says Hennessey. “It should be feasible, and if we don’t achieve it we should be ashamed of ourselves.”

More than just a toy

Earlier this year, smart-toys company Elemental Path unveiled its first product, the CogniToys Dino. With Watson and the company’s own Friendgine platform, the small toy dinosaur connects to the internet to interact with children, answering their questions, telling them jokes and playing games.

But it’s not a one-size-fits-all product. The connected nature of the toy allows it to learn about a child’s preferences and interests, which results in a more personalised experience each time the two interact. “It’s a lot like a kid-friendly, educational version of Siri in the palm of a child’s hand,” says Elemental Path chief executive Donald Coolidge, who co-founded the company in 2014 after he and a team were crowned one of three grand prize winners at IBM’s Mobile Developer Challenge.

Coolidge, who has created apps and websites for toy companies throughout his career, said having the chance to start Elemental Path served as a tipping point for him as he “really did not love the way the toy industry was using technology”, and thought brands were relying too heavily on mobile games and apps.

“I wanted to take that two steps further and not just create an app that went along with a toy product, but really jam the technology right into the toy and make it the foundation of the product,” he says. Feeling that kids already spend too much time staring at phones and tablets, he was intent on creating a digital toy that didn’t include a screen.

That led to the creation of the CogniToys Dino. To answer children’s questions, the Dino digs into a dictionary full of kid-friendly definitions that Elemental Path has uploaded into Watson. So if a child asks its Dino what a scientist is, the toy will provide an answer that is digestible and age-appropriate, meaning that a five-year-old will get a simpler explanation than a nine-year-old.

“Watson allows us to ask a database that has a lot of information a question in a natural way, and returns an appropriate answer,” Coolidge says.

Coolidge adds that Dino is just the beginning for CogniToys and that he hopes to partner with well-known toy brands down the line to help them integrate the technology into their own products. “We’re not a toy company, we’re a technology company,” he says. “We see ourselves partnering with big entertainment and toy companies that have really established brands with great content, and empowering those products with speech and personality.”

Elemental Path is also focusing its efforts on CogniToys’ potential to help hospitalised children. The company is preparing to conduct a pilot in partnership with a hospital to determine whether it can help reduce stress or anxiety levels in child patients. According to Coolidge, toys like the Dino could help comfort hospitalised children by answering questions that they might be too afraid to ask a doctor or nurse, such as ‘what is cancer?’ or ‘what is a biopsy?’

Yet for all the good that connected toys could provide in the realms of education and healthcare, there have been some concerns among parents regarding privacy and security. “For us, it’s really important that we actually understand the kids using the toy, because then we’re able to give them a better, more educational and more personalised experience,” Coolidge says. “But we understand that the end user of the product is a five to nine-year-old, so it’s important that we take as many steps as possible to make sure we’re collecting only the information that is necessary.”

Coolidge says that when parents download the app to activate their child’s CogniToys Dino, they are not asked for any personally identifiable information like a credit card number or home address. Instead, he says, the app asks questions about the child so the Dino can have a base to start with, like ‘what type of foods do they like?’ and ‘what sports do they like?’

Another, albeit less severe, argument is that these kinds of toys could potentially stifle imagination as they allow so much information to be readily available at a child’s fingertips. But Coolidge argues that toys like the Dino actually encourage a child to use his or her imagination since they aren’t being forced to take part in anything that’s pre-programmed.

“We really feel like kids can have a deeper type of engagement with the toy,” he says. “Kids will say stuff like ‘well, I have to go to school now, I need to turn you off’ or ‘I need to go to bed now, goodnight’ and it’s really cute, but it does show a deeper level of engagement, where the toy becomes more than just a toy.”

This article was first published in The Drum's AI issue, guest edited using IBM Watson technology.