IBM Watson AI criticised after giving 'unsafe' cancer treatment advice

Internal IBM documents reveal criticisms of its supercomputer's diagnostic ability - AP
Internal IBM documents reveal criticisms of its supercomputer's diagnostic ability - AP

IBM's Watson supercomputer has come under fire for providing incorrect and 'unsafe' healthcare treatment advice to cancer patients.

The system is being used in 230 hospitals around the world to help doctors diagnose patients. It does so by using artificial intelligence to analyse their medical data in combination with information from hundreds of medical journals. Since 2015, Watson has given advice on nearly 60,000 patients.

A report from health website Stat News states that internal documents shared by IBM Watson's former deputy health chief Andrew Norden provided strong criticism of the Watson for Oncology system.

It stated that the "often inaccurate" suggestions made by the product raise "serious questions about the process for building content and the underlying technology".

For example, Watson reportedly suggested giving the drug Bevacizumab to a 65-year-old man diagnosed with lung cancer, who also seemed to have severe bleeding. One of the side effects of the drug is that can lead to “severe or fatal hemorrhage".

According to the documents reviewed by Stat, a doctor at Florida's Jupiter Hospital told IBM: “We bought it for marketing and with hopes that you would achieve the vision. We can’t use it for most cases.”

A healthcare revolution: how the NHS wants to use technology to keep you out of its hospitals
A healthcare revolution: how the NHS wants to use technology to keep you out of its hospitals

According to the report, the documents blame training provided by IBM's engineers and doctors from a private clinic, the Memorial Sloan Kettering Cancer Centre in New York, which had partnered with IBM on a project to adapt Watson to the healthcare sector.

IBM Watson gained worldwide fame when it won American gameshow Jeopardy in 2011 and was adapted to be used in different sectors.

In healthcare, Watson was adapted to assist in diagnoses and treatment of patients, identifying key pieces of data through the description of symptoms and mining patient data to find relevant facts about family history, current medications and other existing conditions.

IBM boosted the capacity of the Watson supercomputer by acquiring medical imaging group Merge Healthcare for $1bn in 2015. 

At the time, IBM said: "The planned acquisition bolsters IBM's strategy to add rich image analytics with deep learning to the Watson Health platform - in effect, advancing Watson beyond natural language and giving it the ability to 'see'."

According to IBM Watson can then provide a list of potential diagnoses along with a score that indicates the level of confidence for each hypothesis.

Children’s hospital Alder Hey in Liverpool turned to IBM’s artificial intelligence system Watson in 2017 for its analytics needs in a bid to become the UK’s first cognitive hospital.

IBM's AI healthcare proposition received a blow last year after the University of Texas MD Anderson Cancer Center put their partnership on hold after four years of research. After four years, the partnership did not produce a tool that was ready to go beyond pilot tests, according to website Technology Review.