A prediction model for Clostridium difficile recurrence

Background: Clostridium difficile infection (CDI) is a growing problem in the community and hospital setting. Its incidence has been on the rise over the past two decades, and it is quickly becoming a major concern for the health care system. High rate of recurrence is one of the major hurdles in the successful treatment of C. difficile infection. There have been few studies that have looked at patterns of recurrence. The studies currently available have shown a number of risk factors associated with C. difficile recurrence (CDR); however, there is little consensus on the impact of most of the identified risk factors. Methods: Our study was a retrospective chart review of 198 patients diagnosed with CDI via Polymerase Chain Reaction (PCR) from January 2009 to Jun 2013. In our study, we decided to use a machine learning algorithm called the Random Forest (RF) to analyze all of the factors proposed to be associated with CDR. This model is capable of making predictions based on a large number of variables, and has outperformed numerous other models and statistical methods. Results: We came up with a model that was able to accurately predict the CDR with a sensitivity of 83.3%, specificity of 63.1%, and area under curve of 82.6%. Like other similar studies that have used the RF model, we also had very impressive results. Conclusions: We hope that in the future, machine learning algorithms, such as the RF, will see a wider application.