How AI could make dashboards redundant within the next 10 years

The summit’s first panel discussed how NHS organisations can scale up AI to improve services, and how big innovations might be closer than many might have thought.

For example, the dashboards commonly relied upon to monitor performance in the NHS today could be made redundant within the next decade thanks to advances in AI, according to Erik Mayer, director of Imperial Clinical Analytics, Research and Evaluation.

Mr Mayer suggested AI could allow people, instead of accessing dashboards, “to be talking into the [electronic patient record] and getting the information back”.

He also referred to the 20,000 free text comments his trust received from outpatients each month, as an example of an area where AI could already be deployed. A machine learning process could sort the comments by themes and tone, into information which staff could use to identify and address problems, he said.

University Hospitals Birmingham Foundation Trust chief medical officer Kiran Patel added many clinicians would welcome the opportunity to optimise what they do and standardise as much as possible. He explained how he had already used AI to reduce did not attend rates in outpatients and to develop a “health equity” tool for elective care, which prioritised based on expected outcomes for patients rather than the length of time waiting.

Dr Patel said he was now looking to use real-time data to enable clinicians to ask questions like “how good was care yesterday” and “how safe was care yesterday?” to “build a Health Observatory”.

Turning to how industry was using AI, IBM Consulting executive partner Michael Conway said AI had completely transformed IBM’s HR department so now “94 per cent of our HR queries by any IBMer around the world are dealt with through automation and only 6 per cent goes to a human being”.

However, The Health Foundation’s director of innovation and improvement Malte Gerhold cautioned taking time to think about what outcomes were trying to be achieved rather than rushing headlong into projects. “We should ask the question ‘What is it we are trying to achieve and what problems we are trying to solve in our organisation – and can AI help us deliver that?’” he said.

Meanwhile, OpenEHR chief executive Rachel Dunscombe cautioned that AI needs lifecycle management and performance monitoring – it was not something which could just be switched on and forgotten about.

She said: “We’re going to have thousands of algorithms. We need to be able to have a cockpit where you see a red light when there is something which is not right.”