Professor Ivana Isgum of the Amsterdam University Medical Centre defines the use of AI in healthcare as “the use of complex algorithms and software to estimate human cognition in the analysis of complicated medical data.”
Speaking at this year’s European CanCer Organisation (ECCO) summit in Brussels, Isgum said, “There are two distinct ways in which AI can be used in cancer care; one is diagnosis and the other is treatment.”
For diagnosis, she highlighted research studies showing how AI was particularly useful in radiology by analysing x-ray photos and CT-scans to find abnormalities and signs of cancer.
RELATED CONTENT
AI was also of use was in analysing and manipulating laboratory research and patient data.
In one medical study, algorithms achieved a better diagnostic performance than a panel of 11 oncologists in a simulated exercise designed to mimic routine pathology work, the conference was told.
However, for Professor Isgum the most exciting potential use of AI in patient care centres around the speed of drug development.
She cautioned against getting too excited about these studies, as there were question marks over the ability to compare the effectiveness of one algorithm against another.
“It [radiology] will change beyond recognition. Everything is already changing in hospitals, except radiology departments, where technicians are still sitting in the dark, examining x-ray photos and CT scans” Professor Regina Beets-Tan, ESR
“Clinical implementation requires dealing with major legal and ethical questions,” she said.
Professor Regina Beets-Tan, ECCO board member, told the audience that with the introduction of AI, “One thing I am sure of, is that there will be no radiologists in the future.”
Beets-Tan who is also the second Vice-President of the European Society of Radiology (ESR), added, “It [radiology] will change beyond recognition. Everything is already changing in hospitals, except radiology departments, where technicians are still sitting in the dark, examining x-ray photos and CT scans.”
Instead she believed that radiology will become a mix of people with technological backgrounds, who will also contribute to clinical decision making.
Dr Adrian Brady, chair of the ESR quality, safety and standards committee, disagreed with Tan’s view that radiologists will not be needed in the future, but accepted the role will change.
“We are being over worked and don’t have the time to talk to patients. Therefore, if one of the benefits of AI is to remove some of the tedious tasks and allow us to focus on meeting patients, we at ESR welcome this.”
Giving an industry perspective, Bristol-Myers Squibb’s Laura McDonald said that though the introduction of digital innovations can bring immense opportunities, there were still a number of challenges that needed to be overcome.
“Collecting data in the real world is infinitely messy and complicated and even though we see increasing digitalisation, we see holes in the data" Laura McDonald, Bristol-Myers Squibb
In order to deploy algorithms in a clinical situation, McDonald believed there were limitations in accessing data.
“Collecting data in the real world is infinitely messy and complicated and even though we see increasing digitalisation, we see holes in the data," she said.
She pointed out that medical records are not always accurate and that biases in data influence the algorithm in a negative way.
Dr William Allum, consultant surgeon at the Royal Marsden Hospital in the UK, said that as far as surgery is concerned, AI is still very much in its “infancy”, but added that in the future AI’s most obvious use is in robotic-assisted surgery.
He said that on top of the questions concerning costs, there was still no way to properly compare robots to humans.
However, Allum conceded that in the long-term AI will be used to minimise invasive surgery.
“I think AI will give us better focus, with better imagery of anatomy, as we will have better accuracy of where we need to operate.”
“We could use also AI, to learn from the data of previous operations,” he said, noting the importance of striking a balance between machine learning while retaining a human approach to treatment.
Programme officer for EU policies at the European Commission’s DG CONNECT, Saila Rinne said that the possibility of using AI in cancer care comes at an exciting moment, “as the EU will be starting two new funding programmes from 2021, where there will be provisions looking at health care.”
Rinne pointed out that the Commission had already published a policy document in 2018 on the digital transformation of health care, covering three key issues: better access to data especially relating to genomes and images; research and innovation including AI solutions and digital tools, and legislation and ethics relating to the use of these innovations.
Rinne encouraged the cancer experts to be involved in EU policy making, saying “My message to you is that there is a lot the Commission is doing concerning AI and health care, in terms of investments and policies, therefore be active and participate in our public consultations.”