Training to become a doctor takes around seven years in the UK, in terms of a five-year university education and two years training as a junior doctor.

After that, you can add on the time it takes to specialise. This is before you even consider all the worksthat happens in the years before university training; studying A-Levels, UKCAT tests, GCSEs – the time adds up.


Once you are an official, registered doctor, you need to ensure you are making accurate diagnoses to help your patients. It is estimated that 160 hours of reading per week is needed to keep up with all the new medical knowledge that is published, and that was back in 2013. Just think how much research has been published in the past four years alone.


It takes years to keep on top of all the new information.


A supercomputer, on the other hand, doesn’t need 10 years+ of training and hours of reading in order to make a diagnosis. Artificial intelligence (AI) like IBM Watson can be fed hundreds of medical articles, process them, and use this information in order to correctly diagnose cancer or discover genes that may cause the neurodegenerative disease ALS.


Industry analysts IDC predict that 30% of providers will use cognitive analytics with patient data by 2018. This shows that AI and machine learning are changing the way the healthcare industry functions — and it’s not something that is going to slow down anytime soon.

AI’s healthcare boom

“AI is the current interest topic among all stakeholders: in tech companies and the tech industry, as well as healthcare, including hospitals and pharmaceutical manufacturers,” says Harpreet Buttar, industry analyst for Healthcare at Frost & Sullivan.


“There has been a tremendous uptake in the number of companies, platforms and collaborators that are trying to introduce this platform. And they’re choosing AI as a decision-making tool.”


Watson for Oncology is an example of this. This tool, built using IBM Watson by Memorial Sloan-Kettering Cancer Centre (MSKCC), is designed to provide medical professionals with access to cancer data and practices to help doctors create individualised cancer diagnoses and treatments for their patients.

“Think of Watson for oncology as a decision support tool.”

In this application of Watson, the computer is fed all the new and existing data relating to cancer: medical papers, textbooks and records. It takes into account all the evidence that has been published and processes it in a way that would be impossible for clinicians to do.


“Think of Watson for oncology as a decision support tool,” explains Thomas Balkizas, IBM Watson Health executive lead for the UK and Ireland. “It’s a tool that helps clinicians look at all the evidence to help them get to the optimal treatment decision for a specific patient.


“Normally about 98% of healthcare providers and professionals ask about three or four people they trust for their opinion. What we’re bringing is evidence curated by professionals at MSKCC, as a cognitive aid to help doctors, asking ‘Have you considered this, have you looked at this piece of evidence,’ when making decisions.”


Cancer was obvious choice for Watson to investigate in healthcare. It’s the second most common cause of death in the US, and in 2012 it was expected that at one in four people would die from cancer.


“It’s a disease that touches everyone. So oncology is a huge focus for is,” says Balkizas.


As well as this, it’s a tricky one. Cancer is not one disease but hundreds of sub-types, each with a different genetic fingerprint. Having Watson on hand to process all the different pieces of data relating to oncology is crucial to help advance the standards of cancer care worldwide.


The system, developed with MSKCC, has recently been adopted by Gachon University Gil Medical Centre, marking Watson’s first deployment in Korea.


In this way, AI is helping a hospital’s workflow, in the way the staff make clinical decisions, as well as in patient-care delivery.


“It’s affecting and helping growth in the market — towards a more value-based care,” explains Buttar.

Healthcare assistants in your smartphone

Smartphones apps are one of the crucial ways AI is impacting on healthcare. Startups are disrupting the industry with doc-bots, apps powered by AI that can complement your healthcare visits.


London-based startup Babylon has been making waves in the sector, with investments from the likes of Mustafya Suleman, the co-founder of DeepMind. The Babylon app connects users with a doctor through FaceTime or Skype, to help them get access to a medical professional without long NHS wait times.


It features an AI triage service, where potential patients can check their symptoms using an accurate bot. An in-house live challenge between Babylon’s AI, a senior A&E nurse and an Oxford-education junior doctor saw the “check a symptom” service prove to be consistently faster and more accurate in triaging patients than its human counterparts.

“This is not replacing doctors or replacing any professionals. It gives patients information at times when they could not, or do not, want to reach out to a doctor or nurse.”

In an interview with the Guardian, the startup’s founder Ali Parsa said: “A machine learns every time a diagnosis happens. A doctor might do 7,000 consultations. The machine will do thousands of times more. Therefore the speed at which it learns and everything it sees increases. So we believe in time we will do this more accurately than a human.”


IBM believes this could be where AI will expand in healthcare. “One of the things we’re doing is the ability to take all the high-quality information that doctors and nurses who look after these patients have deemed suitable, and give patients access to this information in the form of a chatbot, an application on your phone,” explains Balkizas.


 “This is not replacing doctors or replacing any professionals. It gives patients information at times when they could not, or do not, want to reach out to a doctor or nurse to ask these questions.”


Apps like this could have wide-spread implications for the whole industry.


“A lot of these questions are from patients that otherwise may have turned up at A&E or been worried sick about something that wasn’t as important. But, by being able to get that information, it will be able to let them take better and more informed decisions,” says Balkizas.

Breaking new ground

With a flurry of startups and companies entering the healthcare space to provide AI diagnoses and healthcare apps, medical imaging is the latest sector to get the intelligence treatment.


According to Buttar, in 2015 Frost & Sulivan registered around 15-20 companies using AI for medical imaging. In 2017, there are almost 350 companies in the space.


“So you can see the growth that is happening in the industry and the kind of services that AI is trying to bring,” he says.


In addition, it is being used to empower drug development, for instance by the UK startup BenevolentAI.


“Current drug discovery is hugely expensive, subject to high rates of failure and takes a very long time. Our solution is to apply our properitary AI and machine learning technology, to access and analyse vast quantities of scientific information, test hypotheses and draw conclusions faster,” says James Chandler, vice president of corporate affairs at BenevolentAI.

“The technology enables previously impossible scientific discoveries by finding connections that would otherwise have been missed.”

The startup developed a platform named the judgement augment cognition system to help scientist make faster discoveries. It takes data from scientific sources, such as papers, patents and clinical trials and uses deep learning and natural language processing to analyse and understand the language in order to build reliable knowledge.


“The technology enables previously impossible scientific discoveries by finding connections that would otherwise have been missed,” says Chandler.


Faster drug development could have huge implications for the industry. Drug development comes at huge losses to pharmaceutical companies, such as waiting for a drug to be ready, the clinical trials and the right to launch the new drug.


According to Buttar, this process can take around 16 years. “There are issues with the money involved – and what if a drug doesn’t work? Then [a company] will have to come out of the market and re-enter. This is where AI can help them.”


As a result, drug discovery and medical imaging is where we will start to see more and more companies entering the industry.

Power to the robots?

It’s important not to run away with all the great things AI can do in the healthcare space, but to remember that this is playing with someone’s personal data.


Last year an investigation by the New Scientist revealed that DeepMind’s partnership with the NHS, which is using AI to improve patient care, meant a lot of patient information was made available to a private company. The information available was completely identifiable; including names of patients, addresses and NHS numbers.


“Information is hard to track,” explains Ron Chrisley, director of the Centre for Cognitive Science at the University of Sussex. “Once it gets out of a certain space, say leaves the NHS, it’s hard to tell where it might end up. Our laws need to catch up with these kinds of technologies.”

“AI is successful because it draws on lots of people’s information – information that is very sensitive and very personal.”

Chrisley believes that the way to ensure patient data is protected is for there to be sufficient regulation and legislation for tech startups and the like to follow.


“This is a general point about technology, but it applies to the medical and AI case,” he says. “AI is successful because it draws on lots of people’s information – information that is very sensitive and very personal.


“We need to re-think about what the possible reach of these things might be.”


Yet as the sector grows, so will the regulation and the industry standards. “Although at the moment AI looks like a big technology, in the coming years it will be democratised and decentralised. It will make it easier for people and businesses to access it — leading to better acceptance and better outcomes,” says Buttar.


Currently, the AI-healthcare sector is concentrated in North America, but this is beginning to open up to more countries – particularly in Europe.


This will lead to a huge growth in the companies and techniques we see entering the space. According to research from Frost & Sullivan, revenues from the use of AI in the healthcare industry will increase from $600m in 2014 to more than $6.6bn in 2021.


“AI could lead to a complete change in research, in a matter of oncology, patients, in genomics, everything,” says Balkizas. “It’s so exciting to be part of something that is at the beginning. We’re just scratching the surface now – the beginning of cognitive computing and the way it is evolving to change healthcare.”

Share this article