GI Outlook

GI Outlook


Sinha S. “Technologies your practice cannot ignore.” Presented at: GI Outlook. Oct. 3, 2020.

Disclosures: Sinha reports no relevant financial disclosures.
October 03, 2020
3 min read

AI in GI ‘will have durable impact on the practice of medicine’


Sinha S. “Technologies your practice cannot ignore.” Presented at: GI Outlook. Oct. 3, 2020.

Disclosures: Sinha reports no relevant financial disclosures.
You've successfully added to your alerts. You will receive an email when new content is published.

Click Here to Manage Email Alerts

We were unable to process your request. Please try again later. If you continue to have this issue please contact

Technology is always evolving and enacting change in the world of medicine.

In his presentation for the American Society for Gastrointestinal Endoscopy’s virtual GI Outlook conference, Sidhartha Sinha, MD, from the division of gastroenterology and hepatology at Stanford Medicine, highlighted advances in technology that will transform the way physicians practice medicine.

“I believe all of these [areas] will have durable impacts on the practice of medicine,” he said. “These, and others, have the potential to address some of the issues we face.”

AI for patient care

Artificial intelligence has been one the major areas for recent study at the intersection of health and technology. Sinha said that research on AI — whether through neural networks or machine learning — has grown in recent years. He said in there were upwards of 12,000 studies that explored artificial intelligence in medicine by 2019, compared with just a few hundred in 2010.

While there are already a number of AI diagnostic technologies already approved by the FDA in specialties like oncology and radiology, none have been approved in GI. However, researchers have explored how the technology might impact aspects of the specialty, including adenoma detection rate and survival analysis for different kinds of gastrointestinal cancers.

“High ADR reduce interval colorectal cancer. We know that,” Sinha said. “The vast majority of AI in GI has been focused on this issue.”

There have been 50 studies on using AI for the analysis of pre-cancerous and malignant lesions, including 48 that focused endoscopy, and a majority of those have looked specifically on colon polyps or cancer. Sinha said they have returned overall positive results with accuracy greater than 80%.

Additionally, separate randomized controlled trials showed that deep learning programs reduced blind spots in esophagogastroduodenoscopy and improved ADR in diagnostic colonoscopy, respectively.

In inflammatory bowel disease, Sinha said a machine learning model that used labs, imaging and endoscopy outperformed 6-TGN levels in predicting response to thiopurines. Other studies showed how AI could predict corticosteroid-free remission and identify patients with Crohn’s disease who might be at risk for disease progression or surgery with greater than 80% accuracy, Sinha said.

Despite the current lack of AI approvals in GI, Sinha said they are in the pipeline.

“Prior to the pandemic, it was anticipated that FDA approval for at least two devices would occur in 2020,” he said. “One for gastric cancer and the other ... for colon pathology.”

AI to combat burnout

Burnout is a critical problem in the health care community, Sinha said, citing recent surveys that revealed that more than half of doctors experienced burnout, nearly 90% reported feeling moderate-to-severe stress, and nearly 60% would not recommend their children follow their footsteps into the field as a career.

“This was prior to COVID-19,” Sinha said. “A recent survey reported higher rates of burnout, and almost 10% of physicians surveyed considered self-harm.”

Though the problem of burnout is complex, Sinha said he and his colleagues are working on a project using AI to optimize electronic medical records and increase the amount of time doctors spend with their patients.

“Physicians spend about twice as much time using EMRs as they do with actual patients,” Sinha said. “The data from trainees has shown that reviewing patient records takes up the majority of this EMR time. The overall goal is to develop a more efficient way to extract data from patient records.”

The AI platform extracts data from unorganized clinical records and places it into more manageable categories. Then, it organizes the information, allowing physicians to pick out exactly what they need. Links to the original documents are included in case more data is needed.

Sinha and colleagues set up a testing interface that allows physicians to review records using standard practices as well as the AI platform. They asked physicians to review patient records and provide standard clinical questions while recording the time and accuracy of each review method.

While the final results of the study are still under review, Sinha said the AI platform saved about 20% of time and was just as accurate as traditional methods. More than 90% of the study’s participants preferred the AI platform over standard record review.

“We hope is this technology allows for improved physician clinic experiences, as well as stronger patient-physician relationships,” Sinha said.

Although he foresees a significant role for AI and other emerging technologies in medicine, Sinha said there will always be a role for human physicians.

“The patient doesn’t want to talk to a machine to learn that something is seriously wrong with her health,” he said. “Technology, including AI, will assist physicians take make better decisions and help patients.”