In 2001 AI was ‘just’ a Steven Spielberg film; in May 2018 it is being described by many as a solution too so many ills within the NHS.
On the 21st of May the Prime Minister provided the NHS with her view on the way Artificial Intelligence could revolutionise the delivery of care for patients with Cancer, Dementia, Diabetes and Heart Disease and by 2030 save 50,000 lives. Grand claims and grand plans and a new direction for government. One that focuses on a digital art of the possible although certainly to leap from paper records in vast wire cages and trolleys as an “ok” solution through to AI as an opportunity for the delivery of care is no mean feat, but a goal we can try to play our part in.
The following day Satya Nadella the Chief Executive Officer of Microsoft gathered CEOs and CIOs from digital business from across the UK to discuss what the team at Microsoft described as “Transformative AI”. The CEO used a quote by Mark Wesiser the prominent scientist of Xerox and the father of the term ubiquitous computing to open his presentation,
The most profound technologies are those that disappear. They weave themselves into the fabric of everyday life until they are indistinguishable from it.
This is where we want our EHR to get to!
The conversation continued to try to deliver the fundamentals in AI. Data is what feeds and teaches AI, it provides the fuel to grow to learn the what and the how.
Collecting more data therefore will educate AI more quickly; the next horizon is to make the nine billion micro-processors that are shipped every year become SMART devices. The micro-processor in your toaster, your alarm clock, your motion sensor light can become part of the data collection capability that will be responsible for our education of AI. The sheer growing size of data is something well documented, the creation of data will have reached a new horizon by 2020 and will look something like the figures below:
- A SMART City – 250 Petabytes a day
- SMART Stadium – 1 Petabytes a day
- SMART Office 150 Gigabytes a day
- A SMART Car 5 Terabytes a day
- Your Home 50 Gigabytes a day
- You 1.5 Gigabytes a day
20 Billion SMART Devices will exist in the world
(8 bits to the byte, 1,024 bytes to the kilobyte, 1,024 kilobytes to the megabyte, 1,024 megabytes to the terabyte and 1,024 terabytes to the petabyte) The average mobile phone now has 128 gigabyte; the first man went to the moon on a computer that had less memory)
So much data to educate the AI of the world, the insights that could be gained are incredible.
The journey from what we know as an IT enabled world to a digital world sees the move from ubiquitous computing to Artificial Intelligence as a pervasive way of life and then on to a world where we live in a multi-sense and multi device experience.
The impact on the relationship between us and technology has evolved in how it is perceived; technology was ‘simply’ a tool, initially as AI evolved it worked for us as a subordinate and as AI evolves still further it will become more of a social peer in how we consider what it can offer us in healthcare. The most common Christmas present in the UK this year was one of the voice activated assistant, people all over the UK are now having chats with Alexa, Siri, Cortana or simply saying Hey Google to find out some fact that just alluded them or to ask for a simple task to be done.
The original concept of distributed computing (or cloud) gives us the ability to create the computer power and data storage that is needed to evolve AI capability. Distributed compute adds IT complexity, it is now our job to find ways to tame the complexity by ensuring consistency and a unification of experiences, this applies more to digital healthcare than any other ‘business’ as we try to utilise digital as a way to standardise the delivery of care as much as we possibly can.
The definition of Artificial Intelligence is said to have been first coined in 1956 in Dartmouth, the journey from this definition now includes the term Machine Learning first applied to algorithms that are trained with data to learn autonomously and more recently since 2010 the term deep learning, where systems are enabled to go off and simply learn beyond a set of specific parameters. The art of clinical practice, the need to have a human touch though is well understood in healthcare. This is why more and more AI in healthcare is referred to as an ability to augment the delivery of care, AI does not deliver a solution to offer less clinicians in the service, what it does is remove the need to have clinical time spent on anything other than patient care, AI offers the opportunity to increase the human touch. A further quote reinforces this in the book The Future Computed;
In a sense, artificial intelligence will be the ultimate tool because it will help us build all possible tools.
Eric Drexler author of Nanosystems: Molecular Machinery Manufacturing and Computation (1992)
The journey to AI in our world is getting quicker. The journey to AI being successful is best measured when the different components of it reach parity with us humans;
- In 2016 AI became able to see to the power of us, and passed the RESNET vision test with 96% able to see 152 layers of complexity.
- In 2017 AI became able to understand speech to the same degree we can the 5.1% switchboard speech recognition test.
- In January 2018 AI was able to read and comprehend to the same degree as a human passing the SQuAD comprehension test with 88.5%.
- In March 2018 AI became better than a human at translation, now able to translate in real time successfully to an MIT measure of 69.9%.
The road to an AI augmented world though is about amplifying human ingenuity; AI can help us with reasoning and allow us to learn and form conclusions from imperfect data. It can now help us with understanding; interpret meanings from data including text, voice and images. It can also now interact with us in seemingly natural ways learning how to offer emotionally intelligent responses. A Chat Bot launched in China now has millions of friends on across multiple social media channels, it has learnt to offer help to its ‘friends’ that are demonstrating symptoms of depression, phoning up friends to wish them good night and offering advice and guidance on sleep patterns but in a very human way.
Gartner have reported that the ‘business opportunity’ associated to AI in 2018 is now worth $1.2 trillion! Suddenly AI is the new Big Data which was the new Cloud Computing, which was the new mobile first. All of these terms have had hype but have all in reality brought a new digital pitch to our business strategies and our lives.
Great Ormond Street Hospital in partnership with UCL is leading the way in AI application into healthcare with several projects delivering startlingly real results.
Project Basecode: Transcribing speech in real time and utilising AI capability to add information to spoken word dictation capture.
Project Heartstone: A device for passing messages, verbal and video to patients of GOSH that may be too young to have their own Smart Phone, the device can be expanded to offer services to children who may be deaf or blind.
Project Fizzyo: Puts in place gamification to the delivery of breathing physiotherapy for children with Cystic Fibrosis and captures the information for the clinical record offering analysis as it goes.
Sensor Fusion: Creates perhaps the most immersive AI elements in healthcare today, recording events throughout the hospital, offering machine learning developed advice and data driven descriptions of events as they occur.
At Leeds Teaching Hospitals Trust we have created a platform in the form of our Electronic Health Record (EHR). With this platform we can now begin to consider how this clinical push for AI and the difference it can make to patients lives and the way we work can be achieved in a carful and considered way.
This digital revolution can make a real impact on Leeds; the patients, clinicians and staff enabling us to provide the care we want to provide following the Leeds Way principles with digital as a supportive backbone.
If you want to know more or have an idea as to how you could help in this area get in touch with us via @DITLeeds