CHICAGO (GenomeWeb) – Two years after its creation, the Program for Human Microbiome Research at Medical University of South Carolina is already having a noticeable impact on research. The longer-term vision, however, is to create safer care environments by improving infection control.
"We have a focus now," according to Alexander Alekseyenko, founding director of the Program for Human Microbiome Research. "The first year was for me to probe and figure out who's here, what are the interests, what are the strengths, and what's missing.
"Most importantly, in that year, we figured out how to do analysis of microbiome specimens locally. We are able to provide that capacity to investigators who are interested in it. We are also able to work with external providers for sequencing microbiomes, but if there is a need to do it locally, we can do it," Alekseyenko said.
He praised the active "extensive, active infection surveillance culture program" at the Charleston, South Carolina, university.
"Every inpatient for our hospital system is subject to the program," Alekseyenko said. On many units, an infection-prevention nurse will visit patients upon admission and take swabs looking for pathogens. Knowing a patient's microbiome composition as soon after admission as possible helps improve analytical models, he said.
"These swabs are then processed through the normal microbiology lab procedures, but a lot of the material remains in surplus for access through this program. These are the materials that we are making available for investigators here and that we are using to characterize the microbiome of our inpatient population," said Alekseyenko.
When he came to Charleston in 2015 from his previous position as a microbiome researcher at New York University, Alekseyenko asked MUSC for funding to establish the Program for Human Microbiome Research as part of his employment deal. The initial investment was to make it a "starter program," he said.
"It's not the ultimate investment that MUSC has made," Alekseyenko said. He said that the College of Medicine is preparing a strategic plan that should include a "much wider" plan for microbiome work. More will be known in January.
At NYU, he was at the Center for Health Informatics and Bioinformatics, where he worked with Martin Blaser, who was funded by the Human Microbiome Project.
NYU's microbiome research program is more mature and much larger than what MUSC has developed in the last two years. "Here, it's more developmental work," Alekseyenko said.
But he has high ambitions.
"My goal is to build this ability to analyze microbiomes in specified populations," Alekseyenko said. "We want to populate [our] research data warehouse with the data from patients in our practice at MUSC Health and we want to use these data for analytics to improve our ability to predict what is going to happen to patients of different phenotypes in the course of treatment. That's where we want to be."
In setting up the Program for Human Microbiome Research, Alekseyenko is breaking down informatics into upstream and downstream categories. Upstream informatics, he explained, is for collecting, storing, and preprocessing data. "And then there is informatics on the downstream end, where you actually use that stored data to extract knowledge," he said.
"We were able to establish the pipeline for upstream informatics analysis here, from the sequencing data to data tables that are suitable for statistical analysis, but that's fairly standard for any place that wants to do microbiome research," Alekseyenko said.
"We are developing a schema for how to store the microbiome data close [in the same relational database] to the patient information in a deidentified way in a research data warehouse. It's a work in progress, but I think we have it in a first version," Alekseyenko said.
It is the downstream part that will be the challenge, though some of the work has begun.
"We can build cohorts based on the microbiome characteristics. We can build cohorts based on the patient phenotypes and look at what kinds of microbiomes they have," according to Alekseyenko.
"The core of the information will come from the medical record," which is more operational than research-focused, Alekseyenko said. MUSC's analytics office then can get this information and build predictive models.
Meantime, Alekseyenko has developed a built a semester-long course on microbiome informatics to teach students how to analyze data. "We're targeting students both in biomedical sciences and statistical sciences," he noted. Alekseyenko also has created an intensive, 2.5-day course for a summer institute at the University of Washington's Department of Biostatistics.
Eventually, MUSC wants to measure the microbiomes of every inpatient in the MUSC Health system, Alekseyenko noted on a panel he moderated at last month's American Medical Informatics Association annual symposium in Washington, D.C. This, Alekseyenko explained, would help the university apply precision medicine to triage and to interventions for infectious diseases.
At the moment, however, the program remains firmly in the research realm.
"We would like to see if the data we can get from these specimens are allowing us to improve the quality and accuracy of our operational analytics in our inpatient population," Alekseyenko said. This includes metrics like length of stay, mortality, and risk of hospital-acquired infection. "We wanted to see if data from the microbiome would allow us to improve these models," Alekseyenko said.
"Honestly, a lot of hospitals aren't even creating these models and some of the models are not existent in our system as well, so we're kind of doing both at the same time, where we're building these initial, predictive models," he explained.
Alekseyenko would like infection control to become more predictive than retrospective.
"For the most part, [infection control programs] exist for reporting the events that have occurred. Most of these programs employ analysts that sit by the computer and retrospectively look over the medical records and say, 'Oh yeah, we have had this reportable hospital-acquired infection event and we have to get more data and report it,'" he said.
"Retrospectively, once you have more than a certain number of events, you may go to the units and see what can be done in order to minimize it."
MUSC is trying to help infection-control nurses prioritize which patients they should see on a given day by introducing predictive analytics based on microbiome research. "You can pour in resources into this and still have infections. To a certain extent, they are unavoidable, but we want to minimize the number of those infections and to use the resources more effectively," Alekseyenko said.