Can A. I. Tutors Help Train Enough Medical Professionals?

By 2030, the global economy will have led to the creation of some 40 million new jobs in the health sector; mostly in middle- and high-income countries. Paradoxically, the WHO stresses that in parallel, the world will be short of up to 9.9 million physicians, nurses and midwives by that year. Moreover, the availability of healthcare workforce is unequally distributed, with some countries having 5 times more doctors than others; or, in some cases, 9 times as many nurses as others.

The imbalances and shortages are predicted to persist in the foreseeable future; with the WHO adding that even the increase in health workers in the past decade is unlikely to be stable and may not be sufficient to attend to the needs of ageing populations. As such, it is becoming clear that we are not training healthcare practitioners in adequate quantities and not fast enough; leaving the impression that we will live in a world with doctor shortages forever.

But could new technologies within the realm of digital health be of some assistance to train medical professionals? New developments, especially in the field of artificial intelligence (A.I.), offer such an indication. So let’s contemplate how helpful A.I. can be in training medical professionals.

Disrupting the training of medical professionals with new technologies

In the digital health era, technology’s assistance in training medical staff is slowly taking a central stage. Mixed reality simulators have been shown to be a viable option to train orthopaedics trainees. Similarly, virtual reality (VR)-trained surgeons have shown a significant boost in their overall performance when compared to traditionally-trained surgeons.?

A new option has recently shown promise in a recent study: artificial intelligence. A recent cognitive assessment led by Dr. Jason Harley from McGill University had 70 medical students perform virtual brain tumour surgery on a simulator. The participants were divided into three groups: one receiving instructions from the Virtual Operative Assistant (VOA), an A.I. tutor to teach safe and efficient surgical techniques and provide personalised feedback; a second group in contact with a remote expert instructor; and a third control group receiving no instruction.

The researchers found that those trained with the A.I. tutor showed 35% better performance and learned surgical skills 2.6 times faster than those receiving remote instructions from a human tutor.

"Intelligent tutoring systems can use a variety of simulation platforms to provide almost unlimited chances for repetitive practice without the constraints imposed by the availability of supervision," said Ali Fazlollahi, the study's first author. "With continued research, increased development, and dissemination of intelligent tutoring systems, we can be better prepared for ever-evolving future challenges."

The A.I. tutor in healthcare

One such recent challenge that Fazlollahi alluded to has been COVID-19. The crisis even led some dentistry students to repeat a year due to insufficient hands-on practice and training. With the importance and advantage of remote solutions and digital technologies highlighted during the pandemic, could A.I. tutors be a plausible solution used to address the shortage of healthcare practitioners?

The Medical Futurist Institute previously analysed A.I.’s potential role in addressing the human resource (HR) crisis in healthcare. The paper concluded that while the technology might not immediately solve this HR issue, it can improve the working environment and conditions of healthcare practitioners. While it’s a relatively new approach, there are already real-life examples of such applications.

Founded by a former NHS surgeon, Virti develops A.I.-powered virtual patients that medical staff and trainees can interact with in interactive scenarios through a phone, desktop or VR headset to hone their soft skills. The software relies on speech recognition, Natural Language Processing (NLP) and ‘narrative branching’ to deliver realistic responses, delivering real-time feedback and cues for the trainee to learn and improve from.

The technology is put to use at the Bristol NHS Foundation Trust to safely train medics and surgeons. The Health Education Center at the University of Texas Medical Branch also adopted the technology to enable students to independently work on their communication and clinical reasoning skills.

“We’ve been working with healthcare organisations for several years, but the pandemic has created really specific challenges that technology is helping to solve,” said Dr. Alex Young, CEO of Virti. “It’s no longer safe or practicable to have 30 medics in a room with an actor, honing their clinical soft-skills. With our virtual patient technology, we’ve created an extremely realistic and repeatable experience that can provide feedback in real-time. This means clinicians and students can continue to learn valuable skills.”

Tutoring with A.I.’s future potentials

Using A.I. tutors to train medical students and staff is still a nascent concept, with Virti and its adopters being early examples put in practice. But as the recent McGill University study showed, this approach holds potential in delivering effective training and it’s generating discourse on the subject.

"Artificially intelligent tutors like the VOA may become a valuable tool in the training of the next generation of neurosurgeons," said Dr. Rolando Del Maestro, the study's senior author. "The VOA significantly improved expertise while fostering an excellent learning environment. Ongoing studies are assessing how in-person instructors and AI-powered intelligent tutors can most effectively be used together to improve the mastery of neurosurgical skills."

This could lead to more interest in the field to develop more A.I. tutors in other fields of healthcare for different purposes. Such a tutor could, for example, help medical students by summarising information coming from traditional resources and subsequently assess their knowledge. Going further, an A.I.-powered system could derive insights from trainees’ wearables to determine when they are at their peak performance to take an exam or perform surgeries successfully.

At the other extreme, such applications could lead to a dystopian scenario where an A.I. system paired with facial recognition technology and cameras constantly monitor students’ behaviour and whether they are displaying signs of learning, aggregating their metrics on a scoreboard in the guise of motivation. Such a scenario unfolded in China and students overwhelmingly disapproved of such a system. As such, the discourse around A.I. tutors should also include such extremes to ensure the ethical use of such tools.

Nevertheless, we might still be years away before seeing such options on the healthcare market, but A.I. offers the potential to attend to, at least partially, the need to train healthcare professionals in crucial skills.