(Doctor + AI) = Saved Lives

 Google

Google

This week, we learned that AI from Google can now spot breast cancer cells more accurately than a human equivalent. Cue a wall of articles about the end of doctors. 

In a blog post, Google explained how LYNA — short for Lymph Node Assistant — can tell the difference between cancerous and non-cancerous cells with 99% accuracy. Staggering.

Working in a confined test environment, LYNA was also able to accurately pinpoint the location of cancers and suspicious regions, some too small to be consistently detected by pathologists. 

Here we go again. Lawyers. Accountants. Doctors. The machines are coming... but, fortunately, the machines still need help. As Google says, with AI improving, LYNA offers "the intriguing potential for assistive technologies to reduce the burden of repetitive identification tasks and to allow more time and energy for pathologists to focus on other, more challenging clinical and diagnostic tasks."

Ian Bremner made a similar point on Twitter, asking what could be better than a 99% success rate? A 99.5% success rate, that’s what, which is what is achieved when doctors and AI work together to detect cancerous cells. 

This is yet more proof for a sustainable model for AI adoption in the near term, with humans overseeing AI, correcting their application, improving their output. 

AI + (Doctors/Lawyers/Accountants) = Better Outcomes For Patients/Clients 

Interesting thought experiment - If you had to be screened for cancer, would you trust a machine? And should you be given a choice? In the UK, with the NHS so stretched, surely the handling of repetitive identification tasks, such as screening, by machines is a positive that would stem a vast cost drain. But would patients be happy about it?

Of course, we should. We trust machines to conduct all sorts of other tasks and medical procedures, including machines that now complete operations alone. But, when it’s come down to it, and it’s my lung or testicle, would I choose to hand full responsibility to a machine?

Machine-only, I’m not so sure. But, for sure, I'd trust (AI + doctor), which is another argument in favour of this collaborative approach. Statistically better results and higher level of reassurance, too. That’s the future.

Edward Playfair