Machine Learning and Robotic Surgery

[GaP42]

Some of you here have developed a fondness for Chat GPT and its capabilities, others have decried the failings providing sensible output and lack of originality.

You might like to consider the capabilities shown by machine learning in conjunction with Chat GPT is this article out of John Hopkins - use of "imitation learning to train surgical robots eliminates the need to program robots with each individual move required during a medical procedure".

Robotic surgery is already here with the da Vinci system - many regular procedures are performed on this remote platform. This is a significant step forward in the programming of the system - "by watching videos of seasoned surgeons, [the system] executed the same surgical procedures as skillfully (sic) as the human doctors."

It was limited to three common task, but they suggest the flexible learning system shows great promise.

LINK:
Surgical Robot Learning

Video of robotic autonomous knot -tying

Trust it?
How confident are you that the learning capabilities cannot be applied to programming? (no, not from video, but how patterns and solutions can be adapted and applied). It may not be too far off.

 

[Jon]

AI will be able to do anything humans do, do it better and it will happen faster than people realise. Just look how fast AI has improved over just the last 3 years. From virtual anonymity to hundreds of millions of users. It was a bit of an iPhone moment.

 

[The_Doc_Man]

I wasn't aware of the AI robotic surgery system, but I was fascinated by an earlier article I read on the web a few years ago, related to a medical project with IBM's "Watson" AI working in conjunction with oncologists to assist in cancer detection and treatment programs. They had a very high success rate, though a few human oncologists could sometimes be more accurate. Oddly enough, a third source of diagnosis for skin cancers turned out to have even better early detection rates than humans... and that was with scent hounds like beagles. Gives a whole new meaning to PET scans.

 

[Isaac]

I am imagining the potential medical malpractice lawsuit and then am imagining the lengthy chain of causation that might be asserted by defendants relating to software, AI, etc. Very interesting. AI will always be a part of our lives, and yes an increasingly big part, but I think the advantages will plateau at the point where it takes more time/knowledge/expertise to massage the input then it would to do the task yourself, that will be the natural factor that people primarily will use to determine whether to or to not use AI for any given task.

 

[AccessBlaster]

I highly recommend the da Vinci system, pretty sure it saved my life a few years back.

 

[Jon]

I highly recommend the da Vinci system, pretty sure it saved my life a few years back.

Isn't that the device that allows for more precise and less invasive surgery? I am curious, were the operators of the device local to where you had surgery, or were they remote in a different location, like 100 miles away or something?

 

[AccessBlaster]

Isn't that the device that allows for more precise and less invasive surgery? I am curious, were the operators of the device local to where you had surgery, or were they remote in a different location, like 100 miles away or something?

My surgery was at UCLA, the surgeon and his team were in the same room as me. He was positioned along one wall as I recall.

 

[Jon]

Ok, so kinda normal sounding. I was thinking if they were 100's of miles away, that would seem a bit mindblowing. I imagine it would feel a lot more comfortable if they were in the same room, in case something went wrong.

 

[GaP42]

Fascinating to see how AI can be applied in the highly professional/expert medical field and how readily embraced it is - driven by cost (patient/ service costs) and time (service/expert availability). Full autonomy may be a while off - da Vinci is generally used in the theatre from a console.

Re litigation: If the AI systems eventually turn out to provide better outcomes then might the higher risk then weigh against the use of human intervention/ involvement? Suing for poor practice/ bad outcomes frequently appears to be an assessment of expert opinions about possibilities not actuals (but then it would no longer require a judgement).

Another aspect associated with this will be the willingness of the "experts" (as the gatekeepers effectively operating as a trade union) to allow / accept the regular use of such robotic systems as best practice, as opposed to their own expertise. We (the public) rely upon experts that have a vested self-interest. Public exposure of observations/assessment/analysis of "outcomes" using a "science-based" medicine approach is an essential part of the process. Thinking about the process for robotic surgery to become mainstream and advance medical practice - not the technology as such - how does any novel surgical procedure become accepted. The key step (for the novel procedure) being the move to perform on actual patients, with ethical approval/ informed consent enrolment in the clinical trial) and to do this multiple times in order to obtain sufficient statistical info conduct analysis to determine clinical efficacy as against gut feel. Hats off to those that take on the task - medics and patients.
[I recall when in discussion with some surgeons providing input to an operating theatre DBMS that it was said that many surgical practices employed did not go through rigorous scientific evaluation when first used - although I would think to a large extent that many would have been subject to review]