Bloodstream infections, particularly those due to antibiotic resistant bacteria, are a major threat to our most vulnerable hospitalized children. Currently there is no good way to determine which children are at high risk of these serious infections. Most of these infections are caused by bacteria or fungi that reside in the patient’s own intestine and therefore are constituents of their intestinal microbiome. Our application proposes to characterize the patient’s microbiome in detail and use advanced computer algorithms to identify patterns in the microbiome that predict infection risk.
Our predictive model will be a completely new and fully functional tool for a personalized risk assessment. We envision a scenario where a patient's sample is sequenced and analyzed within hours of their hospital admittance and then at intervals throughout the patient’s hospital stay. Based on the abundance of potentially harmful bacteria and the presence of antibiotic resistance genes, the patient would be risk-stratified according to their likelihood of developing an antibiotic-resistant infection. Because this technique has the potential for rapid turnaround time, this type of test could be used in a first line of defense for informing clinical decisions about the need for patient isolation, antibiotic treatments, and further testing. In the era of rising antibiotic resistance, such a tool has the potential to dramatically improve infectious disease surveillance and prevent serious bloodstream infections.