AI can save lives, but only if we address data security and usability

The UK government pledged in May to use artificial intelligence (AI) to prevent more than 20,000 cancer-related deaths a year by 2033, with the prime minister’s plans being welcomed by health charities and industry alike.

But with the world still reeling from social media data breaches, will the public be open to sharing the most private of information on the scale needed for Theresa May’s vision to become a reality? And even if they are, will the information be in any fit state to be shared?

On the face of it, the NHS is the perfect place for AI and big data to come together to revolutionise the way diseases are diagnosed and treated.

Every day, the health service has access to reams of the 21st century’s most precious commodity – data. From test results and treatment histories, to lifestyle information and symptom diaries, the list is endless.

Using computers to scan these bottomless pits of information for clues and patterns seems to be a ‘no brainer’ - on the face of it, the NHS has the key to every health-related question we could ever possibly want to ask. But there are two major barriers – data privacy and data usability.

May’s vision of healthcare AI for the UK

As Theresa May said during her speech two months ago, a marriage of AI and big data could save lives.

“Late diagnosis of otherwise treatable illnesses is one of the biggest causes of avoidable deaths,” she said, adding that AI could aid in the early diagnosis of lung, prostate, ovarian and bowel cancer in at least 50,000 people every year.

“The development of smart technologies to analyse great quantities of data quickly, and with a higher degree of accuracy than is possible by human beings, opens up a whole new field of medical research and gives us a new weapon in our armoury in the fight against disease.”

The UK’s healthcare community has responded positively to this call for collaboration between the NHS, advocacy groups and the AI sector to pool its knowledge resources.

Mike Thompson, chief executive of the Association of the British Pharmaceutical Industry, said the plans sent an important message about the need to improve cancer outcomes in the NHS.

“Our research shows that the UK lags behind the rest of Europe for five-year survival rates for many common cancers and is the worst in the EU for lung cancer,” he said.

“Improving rates of early diagnosis will undoubtedly improve the outlook for cancer patients in the UK.”

Sir Harpal Kumar, CEO of Cancer Research UK, agreed, saying that earlier detection and diagnosis could “fundamentally transform” outcomes for people with cancer.

Going digital with healthcare

For real progress to be made, though, the health service needs the infrastructure to utilise the data it has access to. And it needs the public’s trust if it wants permission to share our personal data with the AI industry.

Currently, many services do not even collect electronic data, and those that do have trouble sharing it with team members based at other sites, let alone external entities.

“Advances in detection technologies depend on the intelligent use of data... We need to ensure we have the right infrastructure, embedded in our health system, to make this possible,” said Kumar.

Thompson agreed that the improved use of health data and AI would be critical to achieving success.

While these infrastructure issues are, arguably, easily fixed with the right investment and commitment, concerns over who has access to health information might be more difficult to overcome.

After the last few months of endless GDPR emails and Facebook public hearings, data privacy has never been more under the microscope.

Cyberattack vulnerabilities with healthcare data

The 2017 cyberattack on the NHS saw more than one in three trusts and 595 GP practices infected with the WannaCry virus. It led to 19,000 appointments being cancelled including 100 that were cancer-related.

A report from the National Audit Office said the Department of Health had been warned of risks to its IT systems a year beforehand, but nothing had been done.

More recently, in April an NHS website hosting data from patient surveys was hacked and the breach was not noticed for several days.

While such stories are isolated incidents, they attract attention and erode public trust.

In the US, where a widespread move to electronic health records is taking place, work is being done to understand the barriers to patients consenting to their data being shared.

A nationwide survey of more than 1,600 adults carried out in 2014 found the factor that had the most impact on people’s willingness to share their information electronically was privacy and concluded that addressing these concerns head-on was paramount.

“By making patients more aware of existing privacy policies and (the) security measures in place, the healthcare providers are creating an environment (in which) the patients are more likely to share their private health information, and therefore still able to achieve cost and error reduction benefits,” said the authors.

AI, healthcare and building trust

Allowing the AI sector to come into the doctor’s office with us holds huge potential, and May’s vision of saving lives, and money, by embracing the future of technology is a laudable one.

Before we can achieve it, however, we need to put the right infrastructure in place and build trust in the way data is held and looked after.

About the author: 

Amanda Barrell is a health and medical education journalist, editor and copywriter. She has worked on projects for pharma, charities and agencies, and has written extensively for patients, healthcare professionals and the general public.