In the trial phase, the virtual assistant, called Dora Care, is designed to deliver safe and empathetic phone calls for people with osteoporosis - a bone-weakening condition that affects 3.5 million people in the UK, particularly women and those over 50. 

Fracture liaison services, which monitor these patients, have struggled with staffing shortages in many regions of the UK, and with fracture treatment costing the NHS an estimated £4.5 billion each year.

The two-year project uses generative AI to hold natural, spoken conversations with patients over the phone.

To ensure the AI does not sound robotic or unsympathetic, linguistics experts from Newcastle University first analysed empathetic behaviours by human clinicians.

Dr Adam Brandt, Senior Lecturer in Applied Linguistics in the School of Education, Communication and Language Sciences, said: "We started with a question of ‘how might a patient experience being empathised with’? To do this, we explored how expert human clinicians display empathy, and we found it’s not only what they say, but how they say it, and also when in a clinical conversation they do this. 

“We then looked at how the same linguistic behaviours could be designed into Dora’s conversations.”

Empathy

Dr Kate Preston, from the University of York’s Centre for Assuring Autonomy, which is also involved in the project, said: “One of the first challenges with this project is can AI technology be empathetic and have the desired effect on a patient who is perhaps feeling vulnerable or unwell. 

“Alongside this issue is the question of how we make the technology safe enough to assist in the work of trained clinicians, providing the appropriate information to allow medical decisions to be made for the best possible care of their patients.”

The trial is underway, following medical and ethical approval, and participation is voluntary, with AI calls being used in addition to rather than as a replacement for patients' regular NHS care.

 

Specialists at York University are designing a safety framework to monitor the technology during the trial. 

Professor Ibrahim Habli, Research Director at the University of York's Centre for Assuring Autonomy, said: "Any development and deployment of AI within a healthcare setting must be backed by robust safety evidence and assurance. 

“We know from other projects in healthcare settings, that safety and trust is key to clinicians feeling confident in deploying this technology, and having a clear understanding of how it can assist them in the decision making process.

“AI systems that have been tried and tested, and indeed do the job they have been tasked with, can still be underutilised if the safety assurance is not clear to those who are using them.”

If successful, the technology could be expanded to help monitor other long-term illnesses, such as diabetes, heart failure, and chronic obstructive pulmonary disease (COPD).

The project is led by technology firm Ufonia, in partnership with the Newcastle University, the University of York, with funding from government agency Innovate UK.

Dr Aisling Higham, the project's chief investigator at Ufonia, said: "We are pushing the boundaries of how empathetic and safe autonomous clinical AI assistants can be used to support patients and clinicians.”

 Press release adapted with thanks to the University of York