Machine learning based on tcr repertoire for diagnosis of colorectal cancer
By analyzing the VJ gene combination characteristics of peripheral blood TCR databases, a CRC diagnostic model was constructed using the random forest algorithm, which overcomes the limitations of existing CRC diagnostic methods and achieves efficient and accurate CRC diagnosis.
Patent Information
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- CITY UNIVERSITY OF HONG KONG
- Filing Date
- 2025-01-23
- Publication Date
- 2026-07-17
AI Technical Summary
Existing diagnostic methods for CRC, such as colonoscopy, fecal occult blood test, and imaging techniques, have risks of perforation, false positive results, and radiation side effects. Furthermore, the ML method based on gut microbiota has a high misdiagnosis rate, making new diagnostic methods urgently needed.
By analyzing the peripheral blood TCR database and utilizing machine learning models, especially the random forest algorithm, a diagnostic model is constructed based on the VJ gene combination characteristics of the TCRβ chain to identify TCR database features related to CRC and achieve accurate diagnosis of CRC.
It achieves high sensitivity and high specificity in CRC diagnosis, with an area under the ROC curve (AUC) of 0.9798 and an accuracy of 95%, which is significantly better than traditional methods and has the potential for non-invasive and efficient diagnosis.
Smart Images
Figure CN122417439A_ABST