Refining the Feasibility of Machine‐Learning‐Based Diagnostic Model Utilizing Gut Microbiota Analysis for Colorectal Cancer Screening
Cancer Medicine · 2025
DOI: 10.1002/cam4.70935 ↗Kurz gefasst
Fachartikel in der Zeitschrift Cancer Medicine, 2025. Verfasst von Shintaro Okumura und 6 weiteren Autor:innen. Beteiligte Einrichtung: Research Institute for Microbial Diseases (RIMD), Osaka University, Suita, Japan. Hauptthemen laut MEDLINE: Colorectal Neoplasms, Machine Learning, Gastrointestinal Microbiome, Early Detection of Cancer. Die Arbeit wurde bisher 2-mal zitiert (Crossref) und nennt 29 Literaturangaben. Der Volltext ist frei zugänglich (Open Access, CC BY 4.0). Gefördert von: Japan Science and Technology Agency, Japan Agency for Medical Research and Development.
Abstract (Originalsprache)
Background
Recently, we developed a colorectal cancer (CRC) diagnostic model based on a machine learning algorithm with gut microbiota analysis. In this study, we evaluated the reproducibility of the diagnostic accuracy of the gut microbiota model, compared the diagnostic accuracy of the gut microbiota model with that of the fecal immunochemical test (FIT), and investigated the practical application potential of the gut microbiota model.
Methods
Fecal samples were collected from both CRC patients and healthy individuals (HI) who underwent FIT. Gut microbiota analysis was performed using the same pipeline as that used in our previous study. Study subjects were diagnosed using the machine‐learning‐based gut microbiota model (ml‐GMM) with the same cut‐off value as in our previous study and by FIT.
Results
The true positive rates of ml‐GMM and FIT were 53.1% and 86.4%, respectively, among 81 CRC patients, whereas the false positive rates among 245 HI cases were 7.3% and 2.4%, respectively. Evaluation of the proportion of either ml‐GMM or FIT being positive revealed a rate of 91.4% among CRC patients (Stage 0/I 78.3%; Stage II, 95.5%; Stage III, 96.6%; stage IV, 100.0%), whereas it was 9.4% among HI. Furthermore, we demonstrated a possible synergistic effect of ml‐GMM with FIT for detection of more CRC patients.
Conclusions
The reproducibility of the diagnostic accuracy of ml‐GMM was confirmed. It was suggested that ml‐GMM in combination with FIT could detect more CRC patients than FIT alone.
Schlagwörter
MeSH-Schlagwörter (MEDLINE)
Stichwörter der Autor:innen
Autor:innen und Institutionen
- Research Institute for Microbial Diseases (RIMD), Osaka University, Suita, Japan.
- Graduate School of Medicine, Kyoto University, Kyoto, Japan.
- The Cancer Institute, Japanese Foundation for Cancer Research (JFCR), Tokyo, Japan.
- Uji-Tokushukai Medical Center, Uji, Japan.
ORCID-Profile: Shintaro Okumura, Eiji Hara
Förderung
- Japan Science and Technology Agency
- Japan Agency for Medical Research and Development
Zitieren
APA
Okumura, S., Konishi, Y., Kitano, T., Matsumoto, T., Obama, K., Nagayama, S., & Hara, E. (2025). Refining the Feasibility of Machine‐Learning‐Based Diagnostic Model Utilizing Gut Microbiota Analysis for Colorectal Cancer Screening. Cancer Medicine. https://doi.org/10.1002/cam4.70935
Vancouver
Okumura S, Konishi Y, Kitano T, Matsumoto T, Obama K, Nagayama S, et al. Refining the Feasibility of Machine‐Learning‐Based Diagnostic Model Utilizing Gut Microbiota Analysis for Colorectal Cancer Screening. Cancer Medicine. 2025. doi:10.1002/cam4.70935
BibTeX
@article{Okumura2025,
author = {Okumura, Shintaro and Konishi, Yusuke and Kitano, Taku and Matsumoto, Tomonori and Obama, Kazutaka and Nagayama, Satoshi and Hara, Eiji},
title = {Refining the Feasibility of Machine‐Learning‐Based Diagnostic Model Utilizing Gut Microbiota Analysis for Colorectal Cancer Screening},
journal = {Cancer Medicine},
year = {2025},
doi = {10.1002/cam4.70935}
}
EUVIMED