Cancer treatment plans generated by artificial intelligence (AI) can dramatically reduce the amount of time patients undergo radiation.
Researchers at The Princess Margaret Cancer Centre in Toronto, Ontario, Canada, compared radiation treatments created by machine learning (ML) with those created by human physicians.
They found that patients spend 60 percent less time undergoing painful radiation with the AI treatment compared to the doctors’ treatment plans.
What’s more, an independent board found nearly 90 percent of all ML-generated treatments in the study were considered clinically acceptable for use and more than 70 percent were actually deemed preferable by physicians.
The study did find that physicians are more hesitant to use AI ML generated treatments, though, despite their effectiveness.
AI and machine generated treatments for cancer cane be quicker and more effective than those developed by humans
‘We have shown that AI can be better than human judgement for curative-intent radiation therapy treatment. In fact, it is amazing that it works so well,’ said Dr Chris McIntosh, chair of Medical Imaging and AI at the Joint Department of Medical Imaging and University of Toronto.
‘A major finding is what happens when you actually deploy it in a clinical setting in comparison to a simulated one.’
The study, published in Nature Medicine on Thursday, was conducted by presenting physicians with different radiation treatment options for specific cases of cancer.
One option was generated by an AI, using ML technology and the other was curated by another physician.
The AI generated treatment was faster, having patients spend 47 hours in radiation compared to 118 hours in the human process – a reduction of 60 percent.
Overall, 89 percent of AI treatments were considered acceptable for use and 72 percent were preferable to the doctors’ plan.
While the physician making the choice often preferred the AI generated treatment, when put in an actual clinical setting, many of the physicians still said they would trust the human curated treatment as they felt safer with a human guide rather than a machine one.
While machine generated treatments might be better, physicians are still more likely to use a human generated treatment as there is still some hesitancy many have to trust the advanced technology
‘Once you put ML-generated treatments in the hands of people who are relying upon it to make real clinical decisions about their patients, that preference towards ML may drop.’ said Dr Tom Purdie, an associate professor at the University of Toronto.
‘There can be a disconnect between what’s happening in a lab-type of setting and a clinical one.’
‘If physicians feel that patient care is at stake, then that may influence their judgement, even though the ML treatments are thoroughly evaluated and validated.’
Physicians who were choosing treatments in a theoretical realm, because their patient had already undergone treatment, were likely to have chosen the AI generated treatment.
Physicians choosing treatment for a patient in the real world, as the treatment they chose would actually be used on a real life person, were much more likely to choose the human generated treatment.
The disparity comes from the natural trust the physicians have in their peers, and the potential fear of the AI not working as intended.
Researchers who led the study believe the results are promising, but there needs to be more work done to convince physicians to use AI generated treatments of their patients, as they can be quicker and more effective.
Artificial intelligence already plays a large role in how cancer is treated.
Programs that use AI are already used in drug development and drug recommendations for patients, as the programs can determine how effective drugs will be against certain cancers.
The programs can also be used to guide doctors treatment decisions, as it can identify target areas and determine the most efficient treatment options available.
The Princess Margaret researchers hope their study will lead to AI generated treatments becoming even more commonplace as time goes on.