Arunachalam, Shivaram

INTRODUCTION: Clostridioides difficile infection (CDI) presents a significant challenge in patients with inflammatory bowel disease (IBD), with high recurrence rates and complications. Predicting recurrent CDI (rCDI) in patients with IBD is crucial for implementing targeted interventions to improve patient outcomes. This study aimed to develop and validate a predictive model (RecurCDI-IBD) using supervised machine learning to identify patients with IBD at high risk of developing rCDI. METHODS: Data were collected from adult patients with IBD diagnosed with CDI between 2013 and 2021. Inclusion criteria included adult patients with a confirmed diagnosis of CDI and a history of IBD. The Gradient Boosting Machine learning model (XGBoost) was used to train a binary classification model. Feature engineering included demographic data (age and sex), clinical data (IBD subtype, medication use, and comorbidities), and laboratory data. The primary outcome was the occurrence of rCDI within 60 days of the initial CDI episode. RESULTS: The RecurCDI-IBD model achieved an accuracy of 80.05% and an Area Under the Curve of 0.88 for predicting rCDI. Key predictive features included IBD subtype, sex, specific medications (such as steroids and anti-TNF agents), and comorbidities (such as chronic pulmonary and renal disease). DISCUSSION: The RecurCDI-IBD model demonstrates good discriminatory ability with balanced precision and recall in identifying patients with IBD at higher risk for rCDI. These findings highlight the potential of data-driven approaches to support clinical risk assessment. Further studies incorporating larger and more diverse cohorts and prospective external validation are needed to confirm generalizability and optimize clinical applicability.

INTRODUCTION: Despite research efforts, predicting Clostridioides difficile incidence and its outcomes remains challenging. The aim of this systematic review was to evaluate the performance of machine learning (ML) models in predicting C. difficile infection (CDI) incidence and complications using clinical data from electronic health records. METHODS: We conducted a comprehensive search of databases (OVID, Embase, MEDLINE ALL, Web of Science, and Scopus) from inception up to September 2023. Studies employing ML techniques for predicting CDI or its complications were included. The primary outcome was the type and performance of ML models assessed using the area under the receiver operating characteristic curve. RESULTS: Twelve retrospective studies that evaluated CDI incidence and/or outcomes were included. The most commonly used ML models were random forest and gradient boosting. The area under the receiver operating characteristic curve ranged from 0.60 to 0.81 for predicting CDI incidence, 0.59 to 0.80 for recurrence, and 0.64 to 0.88 for predicting complications. Advanced ML models demonstrated similar performance to traditional logistic regression. However, there was notable heterogeneity in defining CDI and the different outcomes, including incidence, recurrence, and complications, and a lack of external validation in most studies. DISCUSSION: ML models show promise in predicting CDI incidence and outcomes. However, the observed heterogeneity in CDI definitions and the lack of real-world validation highlight challenges in clinical implementation. Future research should focus on external validation and the use of standardized definitions across studies.