Google DeepMind has formed a third NHS partnership as it seeks whether its machine learning technology can be used to treat head and neck cancers.
The AI research lab has teamed up with University College London Hospital for the latest project after previously teaming up with the Health Service on two other projects in the capital.
In treating head and neck cancers, clinicians must painstakingly “segment” cancerous cells from those which are healthy. This information is then fed into a machine which kills off the harmful cells.
The precise outlines required can take up to four hours to construct but DeepMind hopes to cut this to around an hour.
“Our collaboration will see us carefully analyse anonymised scans from up to seven hundred former patients at UCLH, to determine the potential for machine learning to make radiotherapy planning more efficient,” it wrote on its website.
It said head and neck cancers affect over 11,000 patients in the UK each year.
DeepMind Technologies was founded in the UK in 2011 by Mustafa Suleyman, Demis Hassabis and Shane Legg and bought and renamed by Google in 2014.
Its first two NHS partnerships – a monitoring app which seeks to detect early signs of acute kidney injury and machine learning project which looks for early signs of eye diseases – were controversial.
The relationship came under the microscope when it was revealed that DeepMind has access to millions of NHS patient records for the kidney project.
It responded by organising four ‘patient engagement forums’ a year at Google’s new London office in King’s Cross, with the aim of achieving greater transparency.
Announcing the head and neck cancer project, the company pledged to treat patient data with the “utmost care and respect”.