Invited Lecture(JGES)
Thu. November 2nd   11:30 - 12:00   Room 12: Kobe International Conference Center Main Hall
Invited Lecture6
 How to use endoscopy database to generate high impact research?
Francis K. L. Chan
The Chinese University of Hong Kong
1. Data privacy / harmonization / quality - The key is always to identify a common reference key which is unique to represent an unidentified patient (e.g. 001, 002), then we need to clean the data based on this key because there will always be duplicated records, invalid records, etc.
2. Data extraction techniques, data processing, and quality control measures - we need to know what is coded and what is not coded properly, because big data research depends on the diagnosis and procedure coding. The concept of sensitivity analysis is to ensure data quality - choose a subgroup where details can be validated by manual checking.
3. Importance of collaborative efforts between clinicians and data scientists - Clinicians provide important clinical questions and critical analysis of the results, while data scientists provide expert advice on the methodology and complex statistical analysis.
4. Examples of successful research projects using endoscopy - Multicenter database registry for endoscopic retrograde cholangiopancreatography: Japan Endoscopic Database Project: https://pubmed.ncbi.nlm.nih.gov/31361923/ Current status of diagnostic and therapeutic colonoscopy in Japan: The Japan Endoscopic Database Project: https://pubmed.ncbi.nlm.nih.gov/33774877/ Therapeutic endoscopy-related GI bleeding and thromboembolic events in patients using warfarin or direct oral anticoagulants: results from a large nationwide database analysis: https://pubmed.ncbi.nlm.nih.gov/28874418/ Risks of post-colonoscopic polypectomy bleeding and thromboembolism with warfarin and direct oral anticoagulants: a population-based analysis: https://pubmed.ncbi.nlm.nih.gov/33619167/
5. Data sharing and integration across multiple endoscopy databases to enhance research collaborations - Here comes to the importance of open-source data for AI development and validity. Now the emphasis is on sharing raw endoscopic photos e.g. AI-colonoscopy models for cross-training and cross-checking.
Page Top