Development and evaluation of a colorectal cancer screening method using machine learning-based gut microbiota...

Area of Science:

• Microbiome research
• Oncology
• Bioinformatics

Background:

• Gut microbiota alterations are linked to colorectal cancer (CRC).
• Machine learning (ML) models analyzing gut bacteria show potential for CRC diagnosis.
• Previous studies often lacked robust validation with independent datasets, raising concerns about reliability and overfitting.

Purpose of the Study:

• To develop and validate a novel ML-based diagnostic model for CRC utilizing gut microbiota.
• To address limitations of previous studies, particularly small sample sizes and inadequate validation.

Main Methods:

• Utilized gut bacterial DNA meta-sequencing analysis.
• Developed a machine learning model to identify CRC based on microbial signatures.
• Validated the model using independent test datasets from diverse geographical locations and sample collection sites.

Main Results:

• The ML-based CRC diagnostic model achieved a substantial increase in true positive rates as CRC stage advanced.
• Achieved over 60% true positive rate for CRC patients beyond Stage II, with a false positive rate of approximately 8%.
• Demonstrated consistent performance across samples from different cities and various colorectal locations, indicating robustness.

Conclusions:

• The developed gut microbiota-based ML model shows high accuracy and reliability for CRC diagnosis.
• The model's consistent performance across diverse sample sets suggests potential for practical, widespread application in CRC screening.
• This approach offers a promising avenue for non-invasive CRC detection and early intervention.