The Use of AI to Improve Quality of OGD

The Chinese University of Hong Kong

Status: Unknown

Conditions: Anemia; Dysphagia; Dyspepsia; Acid Reflux

Treatments: Diagnostic Test: The use of artificial intelligence system Cerebro

Study type: Observational

Funder types: Other

Identifiers

NCT04883567

2021.061

Details and patient eligibility

About

Early and accurate recognition of gastrointestinal lesions has implications for public health. The quality of oesophagogastroduodenoscopy (OGD) has great impact on the detection of oesophageal and gastric malignancies. An AI system, Cerebro, has been developed as a real-time recognition of a pre-defined 28 locations. This system would alert the endoscopist if the sites were not inspected with adequate time. The aim of the study is to investigate the use of AI in endoscopists-in-training during OGD to provide a real-time feedback for ensuring compliance to a standardized protocol for examination. The hypothesis is that the use of AI system Cerebro will significantly improve the compliance to the protocol and thereby reduce the blind spot rates of OGD, ensure adequate inspection time and ensure adequate photodocumentation.

Enrollment: 210 estimated patients

Sex: All

Ages: 18+ years old

Volunteers: No Healthy Volunteers

Inclusion criteria

1. All patients aged >= 18 years old
2. Undergoing diagnostic OGD for evaluation of their symptoms in the Prince of Wales Hospital will be included in the study.

Exclusion criteria

1. Patients in which a full endoscopic examination is not required including patients with a specific diagnostic purpose without the need for a full evaluation, follow-up endoscopy within 8 weeks for a previously diagnosed disease or condition,
2. Patients' condition requiring therapeutic endoscopy,
3. Early termination of endoscopy due to patient intolerance, presence of large amount of food residue, presence of mechanical obstruction or for reasons of safety,
4. Patients with altered anatomy such as history of esophagectomy, gastrectomy, gastric bypass surgery, or Whipple's operation,
5. Patient who are not suitable for EGD examination (such as acute peritonitis with suspected perforated bowel)
6. Pregnant females
7. Patients who refuse to participate in the study
8. Patients who are unfit for consent.

Trial design

210 participants in 2 patient groups

AI arm

Description:

AI would be used to assist the trainee endoscopist

Treatment:

Diagnostic Test: The use of artificial intelligence system Cerebro

Non-AI arm

Description:

Trainee endoscopist would perform OGD as usual.