This article is more than 5 years old. Information may no longer be current.
It’s Time for GI to Embrace AI
Click Here to Manage Email Alerts
Artificial intelligence has made it possible for us to achieve the extreme advancements we have recently seen in health care.
For example, studies have demonstrated that the Apple Watch has the ability to detect irregular heartbeat patterns indicative of atrial fibrillation. AI has shown so much promise in health care, that some major health systems are implementing AI into everyday practice.
I firmly believe that AI is changing how we practice medicine and I think it’s critical that we as gastroenterologists take ownership of the application of AI to our field.
Together with Michael Wallace, MD, from the Jacksonville Mayo Clinic, at the first Global Gastroenterology and Artificial Intelligence Summit, which happened last month in Washington, D.C., brought representatives of technology giants like Amazon, Facebook and data scientists, computer vision experts, the FDA and NIH together along with GI and endoscopy experts to bring AI in GI to the forefront.
One of the main themes of the summit was that digital health is here to stay, and we need to provide direction for application in gastroenterology.
There are several ways that AI will impact and can help advance the field of endoscopy and gastroenterology. First is in how we provide education and training to our future GI trainees. We can use virtual and augmented reality to train the next generation of gastroenterologists by using VR and AR images of the GI tract, how to perform endoscopy, and techniques of both diagnostic and therapeutic endoscopy.
Computer vision is another piece of the puzzle. During endoscopy, AI can help aid us in the detection of polyps and neoplastic lesions throughout the GI tract. For instance, we know that there is a miss rate for polyps during colonoscopy, or that some of the lesions may be so subtle, that our eyes have not been trained to recognize them.
Further, machine learning (ML) algorithms will help in improving diagnosis, management and quality of care that we provide to our patients. For example, we currently use adenoma detection rate as a quality measure in colonoscopy. We are always asking our ancillary staff to pull up our reports and calculate the ADR and other quality measures. This leads to the need for additional resources and the possibility of human error. But, AI can automatically conduct all of this. ML algorithms can be applied for prognostication, response to therapy for IBD, risk scoring for GI bleeding, risk for early onset colon cancer and more.
Many advances in AI are here today, some will be here in a few years and others may take several years to come to fruition. Advances in AI in a few years will lead to improvement in our ability to detect more polyps, flat lesions, and indicate the completeness of resection of neoplastic areas. While the procedure is being performed, there’s a screen that will accurately highlight the polyps and neoplastic areas. In a few years we should also expect “smart scribes” writing and printing our endoscopy notes. AI will also provide us with the quality of the bowel preparation during colonoscopy, download and annotate pictures at the same time.
Further down the line there will be prediction and risk stratification models that will be validated. These models will help identify patients at high risk for GI cancers, prioritize admissions for GI bleeding, response to IBD therapy and more. Additionally, ML algorithms will be to be able to scan through patient reports and images and provide a surveillance interval for example, that this patient needs to have a repeat endoscopy in 3 years, and in 3 years send a reminder to the patient and to the primary care physician that the procedure is due.
Although several of these developments are likely later down the road, a lot of research and development is happening across several fronts and it’s just a matter of certain things being available today vs. in the near future.
– Prateek Sharma, MD, FASGE, FACG, FACP
Professor of Medicine
Director of Fellowship Training
University of Kansas School of Medicine
Disclosure: Sharma reports no relevant financial disclosures.
Collapse
Click Here to Manage Email Alerts