U-M researchers launch fight against C. difficile with $9.2M grant from NIH
Prof. Wiens will continue to use machine learning techniques to study the disease.
A team of U-M researchers, including Prof. Jenna Wiens, has been awarded a $9.2 million grant to tackle Clostridium difficile. Clostridium difficile, also called C. difficile or C. diff, is a bacterium that can cause symptoms ranging from diarrhea to life-threatening inflammation of the colon. C. difficile is difficult to eradicate and is often transmitted to patients in hospital environments.
The researchers were awarded the grant from the National Institutes of Health as a government backed effort to attack antibiotic resistant bacteria. They will spend the next five years studying this pathogen that kills over 14,000 each year.
In 2014, Prof. Jenna Wiens led a project that sought to develop and validate a data-driven hospital-specific model for estimating the probability that an inpatient will test positive for C. difficile infection using patient electronic health record (EHR) data and machine learning techniques. EHR data allows for straightforward integration into the health information system and the automatic calculation of patient risk (Open Forum Infectious Diseases).
In contrast to previous risk-stratification models for C. difficile, Prof. Wiens’ project did not limit itself to the set of known risk factors, but considered over 10,000 variables automatically extracted from EMR data. Using machine learning techniques, she developed the model on admissions from a single year and validated it on a held out set of admissions from the following year. The held out set consists of admissions withheld from the construction of the predictive model. The team then compared her proposed model to one based on a small set of known risk variables.
The model that made use of the additional EMR data resulted in fewer false positives and significantly outperformed the model that considered only the small set of known clinical risk factors.
Prof. Wiens will continue to use machine learning techniques to study the disease, and over the next five years the researchers, including team leaders Vincent Young M.D., Ph.D. and Patrick Schloss, Ph.D. (Microbiology and Immunology), will use patient record information and genomic data pertaining to the host and pathogen to get a better understanding of how patients become colonized and infected with the disease. They will also perform tests on laboratory mice to better study the risk factors associated with making patients susceptibility to the infection.
Their findings will be communicated to medical physicians to help prevent C. difficile infections and treat those who are already infected.