Use of AI in Gastroenterology Can Move Beyond “Cool Tools” to Improve Practice Efficiency

66f7e7d30b318b14595dee44ee4f743008ca1f25 1200x681

As artificial intelligence (AI) technology in the gastrointestinal field continues to advance, speakers at Digestive Disease Week 2022 discussed how these tools can be put into practice to improve efficiency, reduce physician burnout and save money.

As artificial intelligence (AI) technology in the gastrointestinal field continues to advance, speakers at Digestive Disease Week 2022 discussed how these tools can be put into practice to improve efficiency, reduce physician burnout and save money.

The session, “Enhancing Your IM Practice with Digital Technologies,” began with an audience member asking moderators and panelists to name the “coolest” recent development in AI. Their responses highlighted that new technologies are emerging every day, but the innovations that really deserve attention are those that will help clinicians and patients. For example, moderator Cadman Leggett, MD, of the Mayo Clinic, noted that new technology can create a “deepfake” image of an esophagus that could trick a gastroenterologist into thinking it’s real – something something that is a cool exploit but has absolutely no clinical use.

In contrast, a presentation by Cesare Hassan, MD, PhD, of Humanitas University in Milan, Italy, provided insight into the use of AI in colonoscopy for automated polyp detection and characterization. He discussed research showing that AI-assisted colonoscopy can halve the rate of missed neoplasms, and argued that even suboptimal machines can be helpful because humans perform far worse.

READ MORE: argues remote meetings can be more productive than in-person

“When you’re distracted, you miss everything,” Hassan said, referring to a human trap that is avoided by machines. And unlike humans, machines can’t lie or cheat, so randomized trials aren’t needed to assess their performance. He noted that AI tools for detecting polyps were incorporated into clinical practice almost immediately, but a paradigm shift is needed for computer-assisted characterization and diagnosis to take hold.

Despite the clear performance benefits of AI for colonoscopy, its cost has prevented widespread implementation in Europe, where Hassan said it’s hard to convince politicians to pay for an expensive tool that will save money over a very long period of time by reducing the incidence of colorectal cancer. . He called for more studies to be conducted on community endoscopy practices, which can help demonstrate the real value of AI tools.

In the United States, the tension is to get insurers to pay for technology that will only yield cost benefits decades later, when beneficiaries will likely have switched to another payer, added the next speaker, Tyler Berzin, MD, of Harvard Medical School and Beth Israel Deaconess Medical Center.

His presentation focused on how AI can help make life easier for gastroenterologists by breaking the cycle of disengagement and burnout that is often accelerated by spending too much time in front of a screen entering data. The exponential growth of patient data and medical knowledge, which he called the “data deluge,” can seem crippling to doctors, and they need something to back them up.

READ MORE:  Delivering learning gains with virtual reality

Enter natural language processing (NLP) and speech recognition, Berzin continued. These tools can derive structure and meaning from language, allowing software to extract data and organize it for analysis. He noted that one opportunity to integrate these tools into the practice of gastroenterology is to generate analyzes for quality measures in a much more efficient way than can be done by humans.

To illustrate this point, Berzin cited research published in Gastrointestinal endoscopy in which researchers developed an NLP algorithm that takes less than 30 minutes to extract data on all colonoscopy procedures performed at their facility since the introduction of electronic health records, whereas manual human review takes approximately 160 hours to extract data for less than 600 patients.1

Berzin also spoke about the potential of workflow solutions to transform the clinician’s experience. These are triage and notification tools that can alert radiologists to which images to prioritize, rather than diagnostic AI tools. These workflow tools may not be as flashy, but they can improve efficiency and present a lower regulatory barrier to approval.

“The goal of AI in medicine is not yet a promise, but it is an opportunity for us to improve clinical knowledge by leveraging data, reduce the quick and superficial work we do and replace it with an opportunity for us to really think like doctors,” Berzin concluded. “I’d bet it would be a very effective way to combat burnout if we could focus back on what brought us to the medicine first.”

READ MORE:  Season 5 Episode 13 Ft. Atom Bomb Body

The final speaker was Prateek Sharma, MD, of the University of Kansas Medical Center, who presented an overview of the future of AI in gastroenterology and where we stand in time. For example, research such as that described by Hassan has moved from conventional endoscopy to computer-assisted detection and diagnosis, but the next steps to take could be automated endotherapy, stand-alone endoscopes, and finally wireless endoscopy. the endoscopist.

Despite the potential of such technology to transform diagnostics, drug discovery and personalized medicine, Sharma said, some key barriers include data access, data security and regulatory issues.

He described the characteristics of responsible AI for the future: repeatable, secure, human-centered, unbiased, justifiable, explainable and monitored.

“Don’t think that’s too much to ask, because it’s the same concept we use for clinical trials of pharmaceuticals, for example,” he reminded the audience.


1. Laique SN, Hayat U, Sarvepalli S, et al. Application of optical character recognition with natural language processing for the extraction of large-scale quality metric data in colonoscopy reports. Gastrointestinal Endosc. 2021;93(3):750-757. doi:10.1016/j.gie.2020.08.038

Source link

Leave a Comment