P0180 - Impact of Artificial Intelligence on Adenoma Detection Rate of Gastroenterologists at a Tertiary Care Endoscopy Suite: A Quality Improvement Study and Descriptive Analysis

Sunday, October 22, 2023

3:30 PM – 7:00 PM PT

Location: Exhibit Hall

Presenting Author(s)

Frances Lee, MD (she/her/hers)

California Pacific Medical Center
San Francisco, CA

Frances Lee, MD¹, Vishwajit Kode, MD², Kassamali Farah, MD², Timothy Chen, MD²
¹California Pacific Medical Center, San Francisco, CA; ²CPMC, San Francisco, CA

Introduction: Artificial intelligence is an emerging tool that allows clinicians to optimize their performance. For gastroenterologists, GI Genius, an artificial intelligence software from MedTronic designed for colonoscopies with computer aided detection (CADe) of polyps, has been approved by the FDA as a medical device. Several randomized controlled trials have observed increases in adenoma detection rates when using CADe during screening colonoscopies. Our quality improvement intervention was to improve adenoma detection rates for screening colonoscopies at our tertiary care center through implementing GI Genius in the endoscopy suite.

Methods: This is a quality improvement and descriptive study seeking to improve adenoma detection rates using CADe at a single center tertiary care facility. The quality improvement intervention was utilizing GI Genius on all screening colonoscopies beginning January 1, 2023. We identified patients undergoing screening colonoscopy from January 1, 2022, to March 31, 2023 using following ICD-10 code “Z12.11.” We excluded colonoscopies with poor bowel prep and those indicated for high-risk features such as prior colon polyp removal, family history of colon cancer, iron deficiency anemia, and positive stool testing. We analyzed polyp and adenoma detection rates before and after CADe integration.

Results: Of the 1,066 colonoscopies extracted by ICD-10 code, 419 colonoscopies were included in our study. 6 gastroenterologists completed 10 or more screening colonoscopies in 2022 and at least 3 screening colonoscopies in 2023. All 6 endoscopists (A-F) met the recommended ADR of 25% in the year 2022 (Table 1, Figure 1). In the first quarter of 2023, 4 of the 6 (67%) endoscopists improved their post-CADe ADR by 1-23 (mean 9.5) percentage points. The highest performing endoscopist with a pre-CADe ADR of 50% experienced a 10 percentage point increase in post-CADe ADR. 2 endoscopists had lower adenoma detection rates, likely due to smaller sample size.

Discussion: Our descriptive study demonstrated that CADe may benefit all endoscopists, even those with high adenoma detection rates. CADe may be an ideal quality improvement measure at endoscopy centers seeking to improve ADR within their facility. With adoption of CADe, future target ADR for standard of care may increase given the potentially large increase in ADR across all practitioners. Future quality improvement investigations may seek to use CADe for improving ADR in trainees.

Figure: Figure 1. Polyp Detection Rate (PDR) and Adenoma Detection Rate (ADR) of 6 gastroenterologists before and after the implementation of CADe on January 1, 2023. A) PDR and ADR for 2022. B) PDR and ADR for 2023. C) Change in PDR and ADR four gastroenterologists. A-F are individual gastroenterologists.

		A	B	C	D	E	F
2022	Colonoscopies with polyp, n	20	9	41	39	17	9
	Colonoscopies with adenomas, n	16	7	31	22	11	6
	Total Number of Colonoscopies, n	33	26	79	56	24	12
	Polyp Detection Rate, %	61%	35%	52%	70%	71%	75%
	Adenoma Detection Rate, %	48%	27%	39%	39%	46%	50%
2023	Colonoscopies with polyp, n	15	1	18	18	1	4
	Colonoscopies with adenomas, n	12	1	12	13	0	3
	Total Number of Colonoscopies, n	23	7	30	21	3	5
	Polyp Detection Rate, %	65%	14%	60%	86%	33%	80%
	Adenoma Detection Rate, %	52%	14%	40%	62%	0%	60%

Table: Table 1. Total number of colonoscopies with polyps or adenomas complete along with Polyp Detection Rate and Adenoma Detection Rate of 5 Gastroenterologists before and after the implementation of CADe on January 1, 2023.

Disclosures:

Frances Lee indicated no relevant financial relationships.

Vishwajit Kode indicated no relevant financial relationships.

Kassamali Farah indicated no relevant financial relationships.

Timothy Chen indicated no relevant financial relationships.

Frances Lee, MD¹, Vishwajit Kode, MD², Kassamali Farah, MD², Timothy Chen, MD². P0180 - Impact of Artificial Intelligence on Adenoma Detection Rate of Gastroenterologists at a Tertiary Care Endoscopy Suite: A Quality Improvement Study and Descriptive Analysis, ACG 2023 Annual Scientific Meeting Abstracts. Vancouver, BC, Canada: American College of Gastroenterology.