X
Business

​AI and machine learning: Building use cases and creating real-life benefits

One organisation, two very different uses of AI and machine learning.
Written by Mark Samuels, Contributor

Getting executives to talk about the potential of much-hyped artificial intelligence (AI) is one thing; getting them to explain how it might work in their business is often another issue altogether.

But Julie Dodd, director of digital transformation and communication at Parkinson's UK, is not only able to talk about how AI might change the game in terms of research and service delivery, she's also able to point to uses cases in her own organisation where emerging technology is already producing life-changing benefits.

Dodd points first to the charity's partnership with BenevolentAI, a specialist in the development and application of artificial intelligence. Parkinson's UK is using their machine-learning platform to search for drugs that might offer a potential cure to the condition. The charity won the partnership in a competition through the Association of Medical Charities.

Applicants were asked to submit proposals that demonstrated how BenevolentAI's technology could solve specific research challenges. Parkinson's UK, together, with The Cure Parkinson's Trust, is using BenevolentAI's capabilities to discover new treatments. Ambitious targets include identifying at least three currently available medicines that can be repurposed to address Parkinson's and two brand-new ways to treat the disease.

The BenevolentAI platform trawls through clinical research data and published studies. It looks at millions of data points and looks for things that humans might not be able to spot, such as potentially positive indicators in otherwise unsuccessful trials. The AI platform comes up with suggested drugs, molecules and pathways that could be beneficial in the future.

"The kind of thing that takes research scientists years, it can do in weeks," says Dodd. "Once we start seeing potential drugs that can be passed through to clinical trials faster than we would have been able to under traditional methods, then that's when digital transformation is going to start making a very real difference to people with Parkinson's."

SEE: How to implement AI and machine learning (ZDNet special report) | Download the report as a PDF (TechRepublic)

Dodd also points to data-led transformation efforts around the charity's advice services. The aim is to use information to help direct the organisation's resources more effectively. She estimates Parkinson's UK can double the amount of people it reaches with its current staffing resources with the help of technology.

Developments have already taken place around user-facing services, and machine learning plays a part here. The charity is developing a product in partnership with other charities that uses machine learning to understand the experiences of individuals with Parkinson's. The tool, which is still at the prototyping stage, analyses the individual, looks at people with similar experiences, and recommends the right advice and support.

"We're looking to avoid the chat bot-type of approach," says Dodd. "We think that form of technology will get there in terms of offering quality advice but, right now, all it does is create bad experiences. We're looking at creating an approach where we provide bite-size chunks of advice that are pulled together by machine-learning algorithms."

The tool collects information via email; people who use Parkinson UK's services provide initial advice about their diagnosis, their personal characteristics, symptoms and challenges. The machine-learning system analyses these responses and the algorithm selects the most appropriate information.

juliedodd.jpg

Dodd: "Digital transformation is going to start making a very real difference to people with Parkinson's."

Image: Parkinson's UK

Those bite-size chunks are then sent to people with Parkinson's via email. Dodd says using email provides the most inclusive approach; the charity's client group tends to skew towards older people, although Parkinson's can affect anyone at any age.

The group working on this prototype includes a range of charities that address other chronic conditions. The technology partner in the project is Reason Digital, a social enterprise that uses its digital expertise to tackle big social issues. Dodd believes the potential long-term benefits are significant.

"What happens beyond this is where things get very interesting," she says. "The machine-learning system works out, from the responses that individuals give, that they have similar challenges to other people. It tells people automatically the kinds of things that other people have found helpful. It learns as it goes, about the things that you like and that you find useful. It's more like the kind of personalised experience you'd find in retailing but it's for people with Parkinson's."

Dodd and her colleagues continue to seek out new applications for emerging technology. While Parkinson's UK is working hard to establish its internal data science capability, the charity recognises that it can fill niche gaps with external expertise.

Take the example of a recent project that Parkinson's UK ran with a data fellow from AI specialist ASI. ASI trains data scientists by letting them work on small but intriguing projects. In the case of its partnership with Parkinson's UK, the data fellow from ASI used machine-learning tools to help the charity make best-guesses about the right peer reviewers for clinical studies.

"That sounds niche but finding the right reviewers to look at your studies is a real problem in clinical research," says Dodd. "And that's something we managed to do in the space of a few weeks using someone with the right data science skills. "

SEE: Sensor'd enterprise: IoT, ML, and big data (ZDNet special report) | Download the report as a PDF (TechRepublic)

Parkinson's UK can already point to some key benefits from dabbling in AI. Dodd says other CIOs who are thinking of taking similar steps into this emerging area must ensure they focus on managing expectations across their organisations.

"People often want to rush ahead," she says. "Unless your data is in a well-structured state, it's very difficult to make the best of AI. People must understand that the quality of data going in is fundamental to the success of the outcomes. You will spend as long structuring your data as you will doing fancy machine-learning projects."

To this end, Dodd has spent time and money ensuring the charity's information is in the right place at the right time. The organisation has implemented cloud-based data warehousing technology from Snowflake Computing to support its digital transformation aims. Dodd says this platform provides a single version of the truth and helps the organisation approach its AI projects with a higher level of confidence.

"Data being accessible is one of the things that people don't talk about very much - but if your data isn't structured, then you can't use it," she says. "Unless you've got a platform that makes it easy to access data, you're not going to make significant progress in AI."

RECENT AND RELATED COVERAGE

How to use cloud computing and big data to support digital transformation

Parkinson's UK is using cloud-based data warehousing to support changes that put better use of information at the heart of its processes.

NHS IT: Using the open standards and the cloud to drive digital transformation

How one of the UK's largest NHS trusts is developing its own infrastructure to put patients at the centre of healthcare.

NHS IT: Can better use of tech give healthcare a shot in the arm?

Is improving tech the key to to making healthcare more efficient?

Wimbledon: How AI, chatbots and big data help serve up a winning experience

How Wimbledon uses cutting edge tech to make sure fans get the most out of the tennis tournament.

Four ways machine learning is evolving, according to Facebook's AI engineering chief (TechRepublic)

Yangqing Jia, director of engineering for Facebook's AI platform team, on the changing field of machine learning.

Alexa, be afraid (CNET)

Commentary: Think Alexa is the voice assistant to beat? Think again.

Editorial standards