Another setback to the UK’s beleaguered National Health Service could point to how artificial intelligence (AI) could help with radiology workloads in future.
The setback, widely reported on in the British press, involved an error in which around 450,000 women failed to get a breast cancer screening reminder. Health secretary Jeremy Hunt said up to 270 women may have died as a result.
Besides making sure such glitches do not crop up again, one of the priorities in any incident such as this is to get screening programmes back on track. In the UK, where radiologists are already stretched, that won’t be easy. However, there are two developments that might help.
The first, which is available today, is teleradiology. Being able to tap into clinicians in other countries, connected via collaborative diagnostic networks, could help national healthcare providers to spread the load when there is an unexpected increase in radiology requirements.
Even teleradiology services, though, might be strained by a mammogram screening backlog, since these usually require full double reading. This is where AI could help, though.
AI tools could, for example, pre-sort exams so radiologists only have to review mammograms where there appear to be signs of a problem. Another option would be for AI systems to fully replace subspecialists for one of the readings.
In the latter case, the profession would effectively move from double reading to single reading with AI support. This is not far off the computer-aided detection used routinely with mammography in the US.
Naturally, for AI to work in this way there needs to be full confidence in its capabilities. It must be allowed by regulators and the tools need to be certified. All this means it may not be an option for some time to come. Nevertheless, the technology offers much promise.
Thus, while AI may never be able to stop mistakes from happening in national health services, it may at least help them to deal with the aftermath.