Too many people, both artificial intelligence (AI) specialists and people who read about the power and potential of AI, make the same mistake many have made over previous technical advances – thinking it’s a panacea. AI is not a solution; it is a tool. It is part of a larger, robust solution. I’m not the only one to say that, but it bears repeating with regularity. I’ve recently had a discussion that helps emphasize that with a cool, real example.
There are many advances in healthcare, and I’ve covered a number of ways that AI can help, ranging from radiology to regulatory compliance and healthcare financial fraud. One area I’ve been watching is the operating theater. I began to talk with one company last year, but they wandered away. On the other hand, I had an interesting talk with Michael Freeman, CEO, Ocutrx Vision Technologies. They are working on improving surgery by combining AI, augmented reality (AR), 5G and other tools to improve both on location and remote (telemedicine) surgery.
An operating theater is a very complex place. One of the best statements of that complexity was created back in the early 1980s, with Monty Python’s “machine that goes ‘ping’.” There are many machines and multiple people working to keep the surgeon informed. She must look in multiple places and at a complex array of information. In the real world surgeons even have health problems with straining to see through machines and at multiple devices. An example of the need to simplify that complexity is the current process of having to look at an MRI and then mentally rotate it to find the right portion of a heart to work on. Misunderstanding is a serious safety issue for the patient.
Initial forays into AR for surgeons has leveraged basic information displayed in a heads-up display, showing health information such as pulse and oxygenation. What Ocutrx and others are trying to do is more complex.
The power of AI allows for 3D rendering of MRIs and the rotation necessary to overlay that image upon the actual heart. The problem is that most AI is still in the cloud, being performed at data centers. For surgery, the connectivity can be too slow. “A surgeon needs to have less than a 10 millisecond delay in response during an operation,” said Michael Freeman. “with the multiple people in an operating theater, all requiring additional information, relying on resources in the cloud is not realistic.”
The company is working on two technical solutions. First, they are moving computing to the edge, to the hospital. Private clouds, or local servers, can provide the scale-out necessary for advanced computing while residing close to operations. Second, 5G is a solution to significantly increase bandwidth, allowing that compute to work during surgery with the short latency required.
“The cloud has been great for developing more powerful computing, but the need for low latency means that on-premises computing is not going away,” said Mr. Freeman. “Lessons learned from the cloud can now be moved to the edge, including combining AI, AR and other technologies to provide a more advanced yet safer operating theater.”
Given that advanced performance, they’re also finding that a key side effect is provided. With people wearing masks and some machines being noisy, relying solely on voice can increase risk. Adding eye tracking to the AR glasses can lower that risk. AI can quickly locate the xyz coordinates of focus, and provide information from certain machines or, combined with voice, send information to other members of the team.
Rural medicine is in an increasingly precarious place in the US, while all over the world there is a lack of specialists. Again, the solution just mentioned can help with that. Imagine an ophthalmic surgeon working on somebody’s eyes. That surgeon has the skills, but not the knowledge. There just aren’t enough specialists to go around and it’s expensive and a strain on the individuals to keep shipping them around.
A strong operating theater solution can be extended to telemedicine. Think back, for instance, to the eye tracking just mentioned. A remote specialist can be looking at an image, and the AI system can identify where the eye is tracking. The surgeon can then communicate an issue directly to the remote specialist who can see the problem area, clearly marked. The system can then take the specialist’s mark-up regions and instruction directly back into the operating theater in real time.
London’s Moorfields Eye Hospital is doing just that. While they have significant knowledge and staff, they’ve also created additional branches throughout the UK and in other nations. While those branches have trained personnel, there’s also a need to link back to the expert knowledge residing in London.
Solutions, Not Tools
As pointed out in the introduction, what this shows is that AI is a tool. Without a doubt, it’s a critical tool. However, it is not working in a void. AI is part of a larger solution and must play well with others. Medicine is not the only example, just an interesting one showing how it takes multiple technologies to be integrated into a solution.
The last few decades saw a move of computing from the edge to the cloud. While massive amounts of compute will remain in the cloud, the advances that have been gained there are helping to move performance back to the edge for critical issues. Hospitals are a place where we will see that migration, and surgery is a place where a combination of technologies will help medical personnel provide better care through providing them critical information in a clear and easy method that helps both physically and mentally – even in the solution doesn’t always go “ping.”