Clinical Evidence

EndoAngel is backed by peer-reviewed research published in the world’s leading gastroenterology journals.

>95%

Polyp Detection
Sensitivity

+15%

ADR Improvement
in RCT

<40ms

Inference
Latency

10+

RCT & Clinical
Studies

EndoAngel can accurately detect focal lesions and diagnose gastric neoplasms by white-light endoscopy

	Focal Lesions		Gastric Neoplasms
	Internal test	External test	Internal test	Prospective test
Accuracy	93.7%	93.3%	88.8%	92.4%
Sensitivity	96.9%	95.6%	92.9%	91.7%
Specificity	90.6%	90.8%	88.0%	92.4%
Positive predictive value	91.1%	92.2%	61.3%	25.2%
Negative predictive value	96.7%	94.8%	98.4%	99.8%

EndoAngel showed sensitivities of 92.9% (internal data) and 91.7% (external) in diagnosing gastric neoplasms. In prospective tests, it detected focal lesions with 92.8% sensitivity and diagnosed neoplasms with 92.4% accuracy, 91.8% sensitivity and 92.4% specificity. Its clinical potential is highlighted in efficiently screening for suspicious lesions and tumors during endoscopy.

Reference: Wu L, Xu M, Jiang X, et al. Real-time artificial intelligence for detecting focal lesions and diagnosing neoplasms of the stomach by white-light endoscopy (with videos). Gastrointest Endosc. 2022 Feb;95(2):269-280.e6. IF = 6.7

Withdrawal time meets standards under ENDOANGEL assistance

The speed monitoring system included in ENDOANGEL ensures more standardized endoscopic procedures: normal speed 0–40, warning speed 40–44, dangerous speed >44. Previous studies reported that the average withdrawal time in colonoscopies was significantly longer with ENDOANGEL assistance (6.38 min vs 4.76 min), ensuring the integrity and standardization of the procedure.

Reference: Gong D, Wu L, Zhang J. Detection of colorectal adenomas with a real-time computer-aided system (ENDOANGEL): a randomised controlled study. The Lancet Gastro & Hepato. Jan22, 2020. IF=30.9

With ENDOANGEL assistance, polyp and adenoma detection rates are improved

Results showed that using the CADe system, adenoma detection rate (ADR) increased from 14.76% to 21.27%, and polyp detection rate (PDR) increased from 41.70% to 55.60%. Using the CAQ system, ADR increased to 24.54%, PDR increased to 53.53%. Based on this study, the CAD system equipped with CAQ raised ADR to 30.6% and PDR to 64.18% without increasing withdrawal time — further ensuring the quality of colonoscopy examinations and improving lesion detection rates.

Reference: Yao L, Zhang L, Liu J, et al. Effect of AI-based quality improvement systems on computer-aided detection efficacy in colonoscopy: a four-group parallel study. Internal review. Nov 25, 2021. IF=11.5

Published Studies

Colonoscopy

Effect of a deep-learning computer-aided detection system on adenoma detection during colonoscopy (ENDOANGEL): a randomised controlled trial

Lancet Gastroenterology & Hepatology / Gut

Prospective, randomized controlled trial — EndoAngel significantly improved adenoma detection rate (ADR) by ~15% vs. standard colonoscopy.

AI quality improvement system for colonoscopy: a four-group parallel study

Multi-center validation study

Multi-center validation of quality improvement capabilities with consistent performance across different operators and clinical settings.

Upper GI (EGD)

Randomised controlled trial of WISENSE for real-time quality improvement of gastric endoscopy

Lancet Gastroenterology & Hepatology

Significantly reduced blind spot rate during EGD. Achieved near-complete coverage of 26 anatomical stations (p<0.001).

Deep neural network improves endoscopic detection of early gastric cancer

Gut

AI-assisted detection showed superior sensitivity for early-stage gastric cancer identification.

EUS (Endoscopic Ultrasound)

Deep-learning-based pancreas segmentation and station recognition system in EUS

Peer-reviewed multi-center study

Validated accuracy for real-time pancreas boundary delineation. Assists both experienced and trainee endoscopists.

Deep-learning-based bile duct annotation and station recognition in EUS