AI Is Set to Revolutionise Healthcare

Doctors have found a new way to improve our health. Alongside traditional X-rays and stethoscopes, they are using data, analytics and artificial intelligence (AI) to help improve patient diagnosis and treatment.

From helping radiologists read scans faster and more accurately to predicting how patients will respond to different treatments, AI is transforming medicine. AI is easing pressure on doctors by facilitating detailed diagnostic work, saving physicians time and improving accuracy. Sophisticated AI software that augments human activities is helping healthcare providers address the challenges of an aging population, the growth in chronic conditions and tightening health budgets.

One field that has been transformed by AI is radiology, an early adopter that has used the technology since the 1990s to analyze scans. Modern high-resolution scanners are producing huge volumes of data, and AI reduces the time needed to interpret these richer and more detailed scans.

New advances in AI are helping radiology departments cope with an increasing workload and a shortage of trained staff. One recent study estimated that the average radiologist interprets an image every three or four seconds, eight hours a day. In the U.K. it’s been reported that the radiology workforce increased by 5 percent between 2012 and 2015. But their workload grew much faster: The number of CT scans in the U.K. increased by 29 percent over the period, and MRI scans were up 26 percent.

To help ease this burden, Siemens Healthineers has developed advanced AI technology to aid radiologists in their work. One example is the visualization software syngo.via which uses AI to detect anatomical structures and to number vertebrae and ribs. Another development is software* which enriches CT images with annotations and data that help radiologists in making a diagnosis.

Deploying AI is part of the larger strategy to use digital technology to cope with the growing demands of modern healthcare. Tight budgets and increasing medical demands have driven the shift to value-based healthcare, which focuses on positive outcomes and measureable improvements in the health of individuals and populations. Creating healthcare value necessitates leveraging the latest technology to improve results. Data from medical devices, wearables, scans and health records is a critical asset that is boosting healthcare, and analytics and AI are vital to interpret this data and turn it into actionable insights that improve diagnosis and treatment.

AI is still at an early stage in many areas of medicine, but it will have a huge impact over the coming decades.

Heart surgeons are employing data and analytics alongside scalpels and stents as they carry out intricate operations, using digital replicas of human hearts and AI to predict the likely outcomes of treatments. In the future, we may all have these replicas—known as digital twins—that are continuously fed data about our bodies and can help predict when we may become ill, and suggest preventive therapy and the most effective treatments. Digital twin technology has the potential to make significant improvements in diagnosis and treatment of a range of conditions.

Building a digital replica of a heart requires collecting reams of data about the patient’s physiological condition, fitness levels and lifestyle.

In one case, cardiologists created a digital version of the heart of a patient suffering from an irregular heartbeat, to test whether the patient was among the 70 percent likely to respond to a particular treatment. When the system confirmed that treatment could be beneficial, the doctors suggested to the replica heart different positions where the electrodes could ideally be placed to get the heart beating regularly again; the digital twin calculated which location would be most effective.

In the long term, digital twin hearts will be useful in reducing the 18 million annual deaths from cardiovascular disease—a third of total deaths worldwide.

“These are fascinating developments that make us optimistic that sometime soon we will be able to address those heavy numbers,” Dorin Comaniciu, Vice President for Artificial Intelligence at Siemens Healthineers, told this year’s Siemens Healthineers Executive Summit, the company’s annual conference for leading healthcare executives, in October.

He explained that the technology behind digital twins is based on a form of artificial intelligence used with neural networks that approximates the workings of human neurons and imitates aspects of human understanding, such as applying predictions. Siemens Healthineers has developed a deep-learning technology, DeepReasoner, that employs multilayer neural networks, allowing data to be aggregated and analyzed in a variety of ways in order to suggest the best decision.

DeepReasoner’s neural networks can be extended to create twins of musculoskeletal systems—the body’s bones, muscles tendons and ligaments. Comaniciu believes that eventually there will be an opportunity for the whole body to be digitally twinned.

“Imagine at some point in the future, our data is being integrated from birth into a lifelong physiological model that follows us through life, so we have neural networks that are constantly analyzing our data, and will expand our thinking from disease care into healthcare. And we’ll talk not about patient-centric prevention, but about person-centric prevention. When disease happens, we’ll have all the information needed to make the right decisions,” he says.

Comaniciu believes that further developments are needed to improve the ability of AI to process large amounts of data, along with better machine learning so that the software can teach itself the most important information about each human body. “Much more has to happen for the digital twin to come to life,” he says.

It is important to remember that, despite the hype, AI is unlikely to replace jobs in healthcare, as some fear. Rather, AI is a valuable tool that improves the speed and precision of physicians by taking over repetitive analytical work and processing vast amounts of clinical data. This frees doctors to spend more time looking at the bigger picture—and caring for patients.

Written by David Benady, for Bloomberg Media Studios

*510(k) pending. This information about the product is preliminary. It is under development, not commercially available, and its future availability cannot be ensured.