Two NHS surgeons use Azure AI to spot patients at higher risk during surgery – Microsoft News Center Brazil

By Andy Trautman

More than 6 million people in England are waiting for treatment from the National Health Service. To put that in context, that’s a million more people than the total population of Ireland.

The COVID-19 pandemic has exacerbated the situation, with staff shortages and non-urgent operations suspended, adding an additional 2.3 million people to waiting lists since May 2020.

The British government is investing 36 billion pounds (about $44 billion) in health and social care over the next three years to “incubate innovation” and end waiting lists. The government said surgical centers, virtual wards and artificial intelligence (AI) “are key to combating backlogs and putting the NHS on a sustainable footing”.

Now, a team of medical professionals from one of the nation’s largest NHS trusts is exploring how artificial intelligence can help reduce wait times, suggest recommendations for healthcare teams and provide patients with better information so they can make decisions about your care.

Orthopedic surgeons Justin Green and Mike Reed of Northumbria Healthcare NHS Foundation Trust have developed an artificial intelligence model that helps consultants provide patients with a personalized assessment of the risks of upcoming hip or knee surgery. This reassures people at one of the most stressful and disturbing times of their lives.

“When I see a patient in the clinic, they look me in the eye and ask, ‘Am I going to be okay?’ “This is very hard to predict, and I end up giving a very general answer,” Reed comments. “I hope this technology will provide a better indication of what is going to happen to these people.”

Justin Green, Leader and Director of Health Education England and an orthopedic intern at the NHS (Image credits: Jonathan Banks)

Green adds that the specialist and patient should always make a decision together about where the operation should take place, and that that decision should be based on the individual’s best interest. But technology could allow them to have a more informed and accurate conversation, while highlighting more relevant information. For example, because the AI ​​model is hosted on Microsoft’s Azure cloud and uses the responsible decision-making dashboard in Azure Machine Learning, medical professionals have a clearer understanding of why AI is coming to these conclusions. This is critical in a highly cautious industry such as healthcare. Counselors can now see how the model works and have confidence that the advice they give patients is based on accurate and reliable data.

The Northumbria Healthcare NHS Foundation Trust is the third largest arthroplasty center in the UK, performing around 3,000 joint surgeries each year. Green and Reed have used the tool in a small number of interactions with patients who need hip and knee surgery, but they believe the concept can be applied to most areas of healthcare.

“I think this will be transformative for predicting the outcome and risks of surgery,” Reed says. “This is just the beginning and there will be many areas that we can explore across health. Not necessary

Being in the field of orthopedics. The concept we developed is fully transferable to predict the risks of any surgical procedure.

We have already had interest from many organizations. The NHS wants to do a pilot because one of its hospitals has 10,000 people on its waiting list. They want to understand if they can offer the surgery in a smaller hospital to individuals at low risk for complications such as strokes and heart attacks. Currently, these patients may have to wait a long time until a larger hospital becomes available.”

A major benefit of Green and Reed’s AI model is that it helps the NHS allocate resources by identifying the specific needs of patients.

Professor Mike Reed, Consultant Traumatologist and Orthopedic Surgeon. (Image credits: Jonathan Banks)

For example, a person in their 60s who does not smoke and who has low blood pressure may be considered a “low risk” patient for surgery. They can have their operation sooner if they decide to have it in a smaller hospital, which may not have the recovery and intensive care areas found in larger hospitals.

Without AI-powered personalized risk assessment, large numbers of patients currently have to wait longer for surgery in large hospitals. This increases the demand for services in large hospitals, as many of these patients can be safely performed in smaller hospitals. In larger hospitals, other critically ill patients can take over the beds, so surgery will likely be called off.

Risk assessment also allows patients at risk of complications to decide whether or not to have the operation.

Green comments: “We consider complications after surgery a bad outcome. It’s expensive for the patient, it’s expensive for the NHS, it can take time and it can prevent someone else from having the surgery. This has a huge impact on the healthcare system as a whole. We can currently give them a risk score.” Generalized saying, “You meet these three criteria, so your risk of a failed trade is 7%, versus the national 2%. “There’s nothing personal about it,” says Green. “As a patient, all I know is that I have three out of seven, my risk is a little high and I can do absolutely nothing about it. So I can decide not to have the operation.”

“We can now show them in detail how the AI ​​model behind this prediction arrives at its findings, which are based on hundreds of data points such as age, blood parameters, body mass index, and past medical history.”

This “how-to” view is only possible because the model runs on Microsoft’s responsible AI dashboard, which helps AI developers with fairness, interpretation, and reliability of AI models. Within the dashboard, tools can communicate with each other and show insights on an interactive dashboard to help with debugging and decision making.

Justin Green and Mike Reed outside Wansbeek General Hospital, which is part of the Northumbria Healthcare NHS Foundation Trust. (Image credits: Jonathan Banks).

Sarah Bird is the Director of the Product Group at Microsoft and leads the responsible and ethical development of Azure AI Cognitive Services.

“The responsible AI dashboard brings together a lot of tools, and this is really useful for an industry like healthcare, which has to ensure that there are no major errors in their AI model and why a particular decision is made,” she says. . “The tools enable teams to control their AI more effectively and help them use it responsibly.”

Mehrnoush Samiki, CTO of Responsible AI Tools at Microsoft, adds that getting a full picture of the ethical principles of AI is critical when it is used in a healthcare environment.

“Azure’s machine learning AI dashboard enables machine learning professionals to train and deploy more transparent, robust, and fair machine learning models across healthcare production cycles,” she explained. “Insights from the dashboard can then be shared through a scorecard, which bridges the gap between machine learning and healthcare professionals, and provides an easy way to communicate insights into model performance and key capabilities that influence patient decision-making.”

Because Green and Reed’s AI model is hosted in the Azure cloud, clinicians can keep track of the entire project and its applications at all times.

The panel could suggest potential data gaps that could give any doctor using AI an incomplete view of a particular patient.

One of the benefits of the Green and Reed AI model is to help the NHS allocate resources by identifying the specific needs of patients. (Image credits: Jonathan Banks).

“Some of Microsoft’s tools for responsible AI are really good and show where these trends are,” says Green. “These paintings are wonderful.”

Reed agrees and adds that having “explainable artificial intelligence” is critical to a healthcare organization.

He also notes that even after several decades of experience in orthopedics, he was surprised by some of the results that the responsible AI committee helped him discover.

“I was looking at what the AI ​​model is looking at to predict the risk of ‘moderate severity’ complications. The dominant factor was age, which was very clear, followed by high blood pressure, which also makes sense. And the third is the platelet count.” These are cells in the blood that help it to clot.”

Reed was surprised to see that platelets play an important role in determining the outcome of surgery when compared to other factors, and this may lead to new areas of research. This finding needs to be validated using different approaches, but it shows how technology helps medical professionals think differently about care.

NHS teams building their own AI models – as Green and Reed have done – are becoming increasingly popular as the healthcare industry tries to manage its growing workloads and provide cutting-edge care to millions of people.

Earlier this year, Health Education England, which supports the delivery of healthcare to the public, published the first roadmap for the use of AI in the NHS, which showed the healthcare industry “recognizing the power and potential of AI to increase resilience, productivity, growth and innovation”.

It is expected that a total of 60 technology professionals will be ready for a large-scale deployment in the healthcare sector in England within a year. There are plans to roll out these and other digital tools in 67 clinical areas, including radiology, cardiology, and general practice.

Patients may not notice changes when they visit their hospital or PA, but they will soon be able to benefit from a more personalized and informative medical experience.

Tags: azure, artificial intelligence, health

Leave a Comment