AI needs quality NHS data to succeed, director of research centre says

For artificial intelligence projects across the country to be successful they need access to good quality NHS data, the director of an AI research centre has said.

Speaking at the 9th National Conference for Radiology Managers Professor, Reza Razavi also said patient engagement is vital if we don’t want to “sour the pitch” for other data-driven initiatives in the future.

Razavi is the director of the London Medical Imaging and AI Centre for Value Based Healthcare, a consortium of academic, NHS and industry partners led by King’s College London and based at St Thomas’ Hospital.

Research teams at the centre are training artificial intelligence algorithms from a large quantity of NHS medical images and patient pathway data to create new healthcare tools.

They include projects including using AI to detect abnormal chest x-rays, predict a patients risk of heart disease and to map out tumours.

But Razavi said the research can only improve health outcomes in the future if it is able to use vast quantities of data within the NHS.

“It’s possible to do, and it’s not to hard to do, if you have access to the data,” he told the audience in London.

“That’s the key thing, having access to the NHS data for reports and imaging.

“And also, if you really want to do the high value things with patients after, then record patient pathways so all the information about them is taken down.”

Alongside quality data, patient perspective and trust is paramount and enough needs to be done to reassure them about how their data is used.

As well as clear discussions on the programme, data opt-out services need to be easily accessible for those who do not wish to participate, Razavi said.

“We are looking at data in a different way. We are very interested in the patient pathway because we think you can learn so much more rather than just seeing images and reports,” he added.

“That’s the key thing here is that the data isn’t leaving the NHS. It’s all managed by the NHS controller, we are just taking the learning out.

“That’s important from the patient perspective as well because patients do worry about their data being sold to a company. We are trying to give them assurances that we are deidentifying that data, nobody knows who it is, and then learning from it.

“Patient public engagement is a really key thing here. If we get this wrong, patients lose confidence and we will sour the pitch for everybody.”

It echoes the same warning issued by NHS Digital chief Sarah Wilkinson at a Kings Fund and IBM Watson event in March, that absolute clarity is needed around how the NHS will use patient data or we risk “deep and almost irreparable mistrust”.

Wilkinson said the NHS needs to formalise its position on the secondary use of patient data in order to fully benefit from the insights already available within the national data set.

She added we need to “directly address people’s concerns by laying out our ethical approach to dealing with data and providing absolutely clarity on how we intend to use health data”.