A study conducted by researchers Julia Powles and Hal Hadison says that an agreement that allowed DeepMind, Google’s artificial intelligence (AI) subsidiary, to access millions of patient records in the UK’s medical system is riddled with errors. The new academic report studying the agreement claims that Mountain View made inexcusable mistakes in terms of oversight and transparency when handling sensitive medical information.
The UK’s data protection watchdog ICO revealed that it is close to resolving the 10-month probe into consent complaints related to medical data-sharing deal between the Alphabet subsidiary and the Royal NHS Trust which operates in three hospitals in London. Tech Crunch notes that the agreement, signed in 2015, had been publicly revealed early last year however, details of how much and which patient data was involved in the deal were not shared.
According to BBC News, over a million patient medical reports were shared with the Google subsidiary to create a software application to alert doctors about patients likely to develop acute kidney injury. The researchers said that it was inexcusable; patients were not aware how their information would be used.
The deal between Google’s DeepMind Technologies and London’s Royal Free Hospital became public last year, causing controversy over the lack of public consultation and the amount of patient detail being shared. Nicole Perrin of medical research charity Wellcome Trust said that it is however, important to remember that fresh data-driven techs do offer promise in improving healthcare.
The study published Thursday in the journal Health and Technology claims that Google’s AI subsidiary made inexcusable errors handling UK health data. The artificial intelligence firm claimed that the reports have major errors. It creates a series of analytical and factual mistakes. The study’s authors said that these accusations are unfounded.
Google’s artificial intelligence company was given access not only to applicable diagnostics and blood tests, but medical records dating back five years, including details on drug overdose, HIV diagnosis, and abortions. According to the Verge, the agreement did not restrain DeepMind from employing AI tech on data.
Powles said that the deal was a very flawed basis on which they initially operated. She added that she is pro data-driven health innovation however, there are real questions of competition, public value, and autonomy that we ought to be looking at. The NHS retains all rights pertaining to make decisions regarding its handling, and the data.
The academic paper heavily criticized the deal between the London NHS trust and the Google-owned company for its lack of transparency. The arrangement between the trust and DeepMind essentially states that the artificial intelligence firm can utilize patient data to build patient portals, medical case management, and the like. The AI firm said that it was commissioning its own rebuttal and analysis, which the study team said they welcomed.
By Anila Maring
Photo Courtesy DeepMind