Gastrointestinal Endoscopy Artificial Intelligence Cloud Platform in Gastrointestinal Endoscopy Screening
- Conditions
- Artificial IntelligenceDiagnoses DiseaseEndoscopyQuality Control
- Interventions
- Device: The Artificial intelligence Cloud Platform
- Registration Number
- NCT05435872
- Lead Sponsor
- Peking Union Medical College Hospital
- Brief Summary
Study objective: To establish a quality control system for gastrointestinal endoscopy based on artificial intelligence technology and an auxiliary diagnosis system that can perform lesion identification, improving the detection rate of early gastrointestinal cancer while standardizing, normalizing, and homogenizing the endoscopic treatment in primary hospitals (including some of the primary hospitals, which are participating in Beijing-Tianjin-Hebei Gastrointestinal Endoscopy Medical Consortium) under Gastrointestinal Endoscopy Artificial Intelligence Cloud Platform as the hardware base.
Study design: This study is a prospective, multi-center, real-world study.
- Detailed Description
This is a prospective, multi-center, real-world study. Before patients are formally enrolled, all endoscopic examination-related systems and endoscopists would be debugged and trained according to uniform standards and requirements, respectively. Patients who meet the inclusion criteria and do not meet the exclusion criteria are enrolled for this trial. All of them will be asked to sign an informed consent after fully understanding the facts about the research study, and will provide demographic information as well as some specific clinical data. Then, participants will be divided into the intervention group (Artificial intelligence Cloud Platform Auxiliary Group) and the control group (Non-Auxiliary Group).
The steps and contents of the gastrointestinal endoscopy examination were completed according to the working routines of the participating units in both groups. Among them, the pre-treatment of endoscopy (such as oral antifoam before gastroscopy, etc. and dregs less diet and intestinal preparation before colonoscopy, etc.) were basically the same in each participating units, and the same equipment and parameters were used to record the whole process of gastrointestinal endoscopy in both groups.
The Artificial Intelligence Cloud Platform in the intervention group can automatically complete quality control, history recognition, and auxiliary diagnosis (an alert box would appear on the display screen to alert the endoscopists) while the gastrointestinal endoscopy process is underway. At the same time, all of the above examination processes would be completed by endoscopists alone in the control group.
After the endoscopists finish writing the gastrointestinal endoscopy reports, the information on desensitized cases will be automatically uploaded to the Cloud Platform database (excluding any sensitive information that may be utilized to identify the patient), including age, gender, examination data, endoscopic examination information (time and pictures), text contents of the report plus quality control indicators. And the pathological results of biopsies during the examination will be added online by the endoscopist when their official reports are released timely.
By comparing and analyzing the results of the two groups, the researchers try to evaluate the performance of the Gastrointestinal Endoscopy Artificial Intelligence Cloud Platform according to the diagnosis rate of early gastrointestinal tract cancer (Primary outcomes) and indicators of quality control of gastrointestinal endoscopy (Secondary outcomes).
Recruitment & Eligibility
- Status
- UNKNOWN
- Sex
- All
- Target Recruitment
- 2000
- From the beginning to the end of the study, patients who received gastroscopy and colonoscopy due to confirmed clinical indications were admitted to Beijing Aerospace General Hospital, Beijing Fangshan District Liangxiang Hospital, People's Hospital of Beijing Daxing District, Gucheng Country Hospital of Hebei Province, and Nanhe Country Hospital of Hebei Province.
- After fully informing and answering the questions, the endoscopic examination with Gastrointestinal Endoscopy Artificial Intelligence Cloud Platform can be accepted, and a signed informed consent form can be provided.
- Patients who refuse to participate in this study;
- Patients with intolerance or contraindications to endoscopic examination, such as severe cardiopulmonary diseases, coagulation disorders, or a total of platelet less than 50*10^9/L.
Study & Design
- Study Type
- INTERVENTIONAL
- Study Design
- PARALLEL
- Arm && Interventions
Group Intervention Description The intervention group (Artificial intelligence Cloud Platform Auxiliary Group) The Artificial intelligence Cloud Platform The patients in this group would be examined by endoscopists with the Artificial intelligence Cloud Platform Auxiliary Device launched with gastrointestinal endoscopy.
- Primary Outcome Measures
Name Time Method Diagnosis rate of early gastrointestinal cancer. two years The number of patients diagnosed with early gastrointestinal cancer is divided by the total number of patients undergoing digestive endoscopy of the intervention group with Artificial Intelligence Cloud Platform Auxiliary and the control group with nothing.
The Early Gastrointestinal cancer in this study is defined as ① early gastric cancer and ② progressive adenoma of the colon and serrated adenoma.
The pathology of biopsies will be referred to the official report of the pathologists in the participating centers, which shall be filled in and uploaded to the cloud platform.
- Secondary Outcome Measures
Name Time Method Indicators for Quality Control of colonoscopy two years The quality control of colonoscopy is assessed with the following criteria: ① Quality of bowel preparations, which is evaluated with the Boston score; ② Withdrawal time, which should be no less than 6 minutes from the time of the first cecum image under colonoscopy to the time of the last rectum image.
Indicators for Quality Control of gastroscopy two years The principle of quality control for gastroscopy in this part is 'no neglected area for observation in the stomach'. The artificial intelligence system can automatically identify the corresponding sites (according to the standard anatomical sites) of the photos taken under the gastroscope and mark them as green on the stomach schematic diagram. After all the sites are observed and corresponding photos are taken, the stomach schematic diagram totally turns green, which would be regarded as no blind sites.
Trial Locations
- Locations (1)
Peking Union Medical College Hospital
🇨🇳Beijing, China