How Is Artificial Intelligence Boosting Clinical Medicine

Artificial intelligence might unleash the real art of medicine. This is how it's boosting medical imaging, diagnosing and procedures to allow healthcare staff to do what they're trained for - caring for patients.

How Is Artificial Intelligence Boosting Clinical Medicine
Photo by jesse orrico / Unsplash

Until now, we have very well established that artificial intelligence is one of the most important players in digital health. We also discussed many areas where it’s applied, but in a very scattered way. The aim of this issue, though, is to look at clinical medicine as a whole and see how artificial intelligence is changing it.

Art

Many say practising medicine is an art. When ancient doctors started documenting their experiences, treating different conditions became more structured. This is when the art got a scientific component. After evidence-based medicine was established, art and science had to be connected. The result became evident in the 20th century, when technology was being increasingly introduced into medical practice. Doctors often consider digital technologies to be a threat to them, but the case is just the opposite. Embracing digital technology and especially artificial intelligence might unleash the real art of medicine.

This is a summary of the abstract of a paper “The Real Era of the Art of Medicine Begins with Artificial Intelligence” published by Bertalan Meskó, the founder of The Medical Futurist.

Technology undoubtedly solved numerous issues of medicine and healthcare. Just imagine still searching for patient data in vast amounts of paper, although, to be fair, that’s still the case sometimes and somewhere. On the other hand, technology presents a certain burden. Instead of just quickly writing a prescription manually, there’s often more administrative work of doing the same digitally. This is still a major contributor to burnout in healthcare staff, which certainly doesn’t contribute to unleashing the real art of medicine.

But what will probably contribute to it in the future is decreasing the amount of monotonous and administrative tasks staff has to complete. The quote from the same paper I mentioned above sums it up perfectly:

”Despite common fears about the perceived dangers of automation, especially artificial intelligence, it does not seem to be taking the jobs of physicians or monopolising medicine. As studies show, it will instead help automate administrative tasks and take over monotonous day-to-day assignments. It has the potential to free up time for medical professionals to let them fulfil the mission they signed up for: to help people on their health care journeys with compassion, creativity, and care.” - The Real Era of the Art of Medicine Begins with Artificial Intelligence

In the next few paragraphs, I’ll discuss three aspects of clinical medicine where AI is increasingly getting adopted - imaging, diagnostics and procedures.

Imaging

Artificial intelligence is already transforming medical imaging in terms of improving image quality and helping interpret these images. Radiology is one area in which technology is becoming ever more present. But that doesn’t mean radiologists will lose their jobs - they will only change how they work. For more, read issue #48.

Why AI Will Not Take Over Radiology
We can often read and hear AI will have the greatest impact on radiology. That radiologists will even be replaced with AI. These were my thoughts exactly, until I dug a little deeper with this article.

However, I recently also discovered that AI is getting adopted in other areas of clinical medicine, such as rheumatology. One example is a company named Capillary, which is developing tools to improve capillaroscopy. This is a non-invasive and inexpensive imaging technique, which is used for example in early diagnosis of scleroderma. Implementing AI will help doctors diagnose it earlier and more accurately. Capillary’s platform, though, is interesting because it’s relatively easy to implement as you don’t need a special capillaroscope. They just offer software that helps to interpret images made with any capillaroscope or microscope available today. Additionally, doctors can collaborate within the program.

I think this is an important point, which they solve efficiently. Instead of medical practices having to change the entire scope of their devices, they just add a piece of software and thus improve it. Like an add-on for your browser. Sustainable and beneficial.

Another area of imaging AI stands to improve is endoscopy. According to a 2020 study in Endoscopy International Open, close to 85% of asked gastroenterologists think a computer-assisted polyp detection (CADe) would improve their endoscopic performance. Additionally, most of them felt comfortable using a computer-aided diagnosis (CADx) system, which would allow them to diagnose cancerous polyps during the endoscopy. The results also show CADx would increase their comfort level when leaving a polyp behind.

Diagnostics

The next area is undoubtedly diagnostics, and the most commonly mentioned application are decision-support systems (DSS). As I wrote in issue #45, DSS have been around for a long time, but the adoption of AI is relatively new. They’re most valuable in analysing large samples of data and suggesting the next steps in treatment.

How Do You Make a Better Decision Support System?
Decision support systems have been around for a long time. But not coupled with AI. A team of researchers is proposing guidelines to make DSS with AI better.

While this section is tightly related to imaging, it’s much more daunting to physicians. If I mentioned how AI could automate administrative tasks and free up time, DSS helps physicians decide about what they’re actually trained for. I agree, DSS seems scary in some cases. But let’s look at it another way.

The difference between humans and AI is that AI has to be very narrowly focused. It needs large amounts of data and a very specific task - this is where it certainly beats humans. But AI doesn’t stand a chance when the broader perspective is in play. Humans are incredibly good at seeing the whole picture - the lab results, patient history and appearance, and integrating all of this into a diagnosis. A DSS suggestion would be just another part for doctors to consider.

For example, a study published in JAMA found out that AI is terrible at detecting sepsis compared to humans. On the other hand, doctors already widely use DSS for managing diabetes, which is beneficial in almost any way you look at it. Patients don’t have to come to the hospitals, and this relieves doctors of a lot of monotonous work. And some of the latest studies show AI might also be used in triaging patients with tuberculosis.

How Can We Use AI to Improve the Diagnosis of Tuberculosis?
There’s a growing number of patients with tuberculosis and the number of radiologists for its diagnosis isn’t following. We will soon need AI for triaging and screening, especially in countries with lower socio-economic status.

Procedures

Lastly, AI also aims to improve medical procedures. A use case I discovered back in November 2020 helps doctors intubate patients. When I first heard about this, I thought to myself, why would one want an AI tool for this and how is it even possible to integrate it into practice. Intubating is pretty critical, and I didn’t imagine when a doctor could find the time to “listen” to an AI system for intubation.

However, when I started reading about it, its usefulness became obvious to me indeed. Based on the report, “a quarter of patients who are intubated outside the operating room have misplaced endotracheal tubes (ETTs) on chest X-rays, which can lead to hyperinflation of the lungs, pneumothorax, cardiac arrest and death”.

The system developed by GE Healthcare gives clinicians feedback on how ETTs are positioned via an X-ray image. Additionally, it’s also capable of detecting the complications caused by misplaced ETTs and alerting clinicians to correct them.

GE Healthcare’s AI tool helps clinicians intubate patients accurately and safely -
An artificial intelligence tool developed by GE Healthcare twinned with a mobile X-ray device can help the placement of endotracheal tubes (ETTs), a necessary step for COVID-19 patients who require ventilation

This is just one example of how AI is used in medical procedures, and in my opinion, such applications are very beneficial. All in all, the benefits are clear on the patient’s as well as the doctor’s side.

After everything written in this post, we can see the benefits of using artificial intelligence in various areas of clinical medicine. I can understand the negative attitude doctors have towards it, I was certainly one of those who thought it was a little scary. But by learning more about it, I changed my perception and realized AI is there to help us, not replace us.