Clinical implications of a machine learning model predicting colorectal polyp recurrence after endoscopic mucosal resection

Abstract

The machine learning model developed by Shi et al for predicting colorectal polyp recurrence after endoscopic mucosal resection represents a significant advancement in the field of clinical gastroenterology. By integrating patient-specific factors, such as age, smoking history, and Helicobacter pylori infection, the eXtreme Gradient Boosting algorithm enables precise personalised colonoscopy follow-up planning and risk assessment. This predictive tool offers substantial benefits by optimising surveillance intervals and directing healthcare resources more efficiently toward high-risk individuals. However, real-world implementation requires consideration of the generalisability of our findings across diverse patient populations and clinician training backgrounds.

Key Words: Colorectal polyps; Machine learning; Risk prediction; Endoscopic mucosal resection; Precision medicine

Core Tip: This predictive model notably enhanced clinical decision-making for colorectal polyp surveillance, demonstrating high accuracy and ease of clinical implementation through a user-friendly online risk calculator. Although promising, its real-world utility depends on external validation, clinician training, and integration with the existing clinical guidelines.

TO THE EDITOR

We read with great interest the recent article by Shi et al[1] in the World Journal of Gastroenterology on the development and validation of a machine-learning-based predictive model for colorectal polyp recurrence one year after endoscopic mucosal resection (EMR). This innovative study addresses a significant clinical challenge, as recurrence rates following EMR remain notably high, ranging from 15% to 50%[2]. The approach proposed by Shi et al[1] significantly improved predictive accuracy, offering valuable advancements in clinical practice.

AI-BASED COLORECTAL POLYP RECURRENCE PREDICTION

The primary strength of this model is the integration of multiple clinically relevant variables including patient age, smoking habits, family history, polyp characteristics, and history of Helicobacter pylori (H. pylori) infection[1,3]. The inclusion of H. pylori infection as a risk factor is particularly noteworthy and is consistent with previous findings regarding its role in gastrointestinal pathology[3]. The eXtreme Gradient Boosting algorithm demonstrates exceptional predictive accuracy, achieving an impressive area under the curve of 0.963 in prospective validation, considerably surpassing conventional predictive methods[1,4]. The model was trained on a dataset comprising 830 patients, with features selected through a supervised process, incorporating clinical relevance and statistical significance.

Clinically, this model can accurately identify high-risk patients, with significant implications. By stratifying recurrence risk, clinicians can tailor follow-up intervals, potentially improving patient outcomes while optimising healthcare resources[4,5]. Nonetheless, it is important to recognise that the model performance may vary across different clinical environments, and external validation in diverse populations is critical to determine its generalisability. Additionally, the interpretability of complex machine-learning models may limit their immediate adoption. Furthermore, the inclusion of an accessible online risk calculator in the model facilitates its immediate clinical application, equipping clinicians with real-time decision support for individualised patient management[1]. However, for successful implementation, clinician training and patient education should be promoted to support the use of these tools in routine practice.

In addition, the integration of this model into existing guidelines, such as those of the European Society of Gastrointestinal Endoscopy, could enhance the personalisation of surveillance strategies.

CONCLUSION

We strongly advocate the integration of this machine-learning-based predictive model in routine clinical practice. This tool demonstrates the potential of precision medicine, refines surveillance strategies, and directs resources more efficiently to improve colorectal cancer prevention and patient care.