Strata Rx 2013: A look at data analytics at Mount Sinai
BOSTON—Mount Sinai Hospital in New York City is on the cutting edge of harnessing data analytics to improve the quality of care at its 1,400-bed teaching facility. Jeff Hammerbacher, founder and chief scientist of Cloudera, and an assistant professor at the Icahn School of Medicine at Mount Sinai, shared some of the organization's data analytics initiatives at Strata Rx 2013.
The medical center relies on computational expertise to bring the tools of analysis to clinical staff and researchers, Hammerbacher said. These tools are intended to enable Mount Sinai to:
- Develop novel therapeutics or diagnostics
- Integrate genomics with the clinic
- Improve outcomes and reduce costs
"We collect structured data, text data like clinical notes, telemetry data, data from our genomics core facility and our imaging core data and integrate them into a single cluster,” he explained. All of it resides in Mount Sinai’s data warehouse and, to date, the data represent 2 million unique patients and 15 million patient visits.
Operational data from its EHR system “is used to treat patients every day,” he said. For example, the hospital has developed a predictive model that analyzes clinical variables and generates the likelihood of 90-day readmissions among patients.
Also, telemetry data is pulled from more than 1,000 beds and used for alerting. Currently, the hospital is looking to integrate this data directly into the EHR to conduct retrospective analyses, he said.
The hospital utilizes multidimensional modeling, or the Kimball model, to analyze 19 dimensions of data that span from 2003 onwards and encompass 500 workflows per night, or 1.6 billion facts—or rows—of data. About 280 researchers query this system 400 times per week to identify a subset of patients for their studies, or develop use cases, Hammerbacher said.
Mount Sinai’s Minerva, a new supercomputer with 30 terabytes of memory, is harnessed to improve scientific discovery for researchers at Sinai. To date, 339 users—including 60 from outside the hospital—tap into the data, which support 161 projects.
Another project, the BioME bank, has enrolled 26,000 patients who have consented to use their data to further research. Of interest, he said, is that each patient also has consented for researchers to contact them in the future. “Often, if you find something outside of scope, you can’t follow up with the patient. But with the BioME bank, you can.”
Mount Sinai’s instruments analyze a tremendous amount of high quality genomics data, he said. The hospital uses data about a patient’s genomics to determine their medication. For example, genetic mutations in a patient inform proper Warfarin dosage. “This is a first-of-its-kind program where it is actually integrated into the EMR, he said.
Also, Mount Sinai is working to turn genome molecules into bits that can be analyzed, and uses visualization of sequences to see variation between certain populations.
The hospital also has made headway with personalized cancer therapy. By analyzing patient profiles and their type of tumor, staff can better identify potential therapeutics and engage in drug repurposing, drug repositioning and immunotherapy, he said.
In one last project mentioned, Hammerbacher said the hospital is working on a predictive tool that analyzes variables associated with hospital acquired infections, which affect 1 out of 20 hospitalized patients.