Data-Based Diagnosis

As overuse of antibiotics gives rise to new generations of resistant bacteria, clinicians face an important decision—hold back the drugs in case the patient turns out to have a viral infection, or prescribe them just in case? Duke engineers are working closely with genome scientists and clinicians to make the decision a snap.

A few years ago, ECE’s Lawrence Carin teamed up with Dr. Geoffrey Ginsburg, professor of medicine and BME and director of Duke MEDx, to combine increasingly efficient gene-expression technology with rapid statistical analysis. A series of DARPA-funded pilot studies proved it possible to read signals of gene expression activity in response to illness before symptoms set in—an advance that could speed up treatment and recovery, and help prevent the spread of disease across populations

Subsequent collaborative projects between the Center for Applied Genomics and Precision Medicine and ECE have improved the methodology as the researchers fine-tuned statistical methods to analyze the data. In a recent study conducted with ECE’s Ricardo Henao, the team proved their technology could rapidly discern the difference between gene expression caused by viruses and bacteria in blood samples—paving the way for a new diagnostic test.

Today, the researchers are working with NIH’s Antibacterial Research Leadership Group and industry partners to translate their innovation into practice, and designing a clinical trial with Duke Health to validate the diagnostic tool in a larger, more diverse population.