Data Driven Care

Illustration by Brennan Wesley and Margaret Dixon Graham

The Big Deal About Big Data

MUSC Health Partners with IBM Watson to Offer Data-Driven Care for Kidney Transplant Patients

By Celia Spell

Hospitals collect reams of data every day, including the number of patients admitted, their vital signs, the length of their stay in recovery—and that’s just the beginning. Processing all of this information can be difficult, so hospitals have started partnering with IT companies to help consolidate and reveal the implications of these data. MUSC Health has decided to harness the power of IBM Watson to do just that. This treasure trove of data can be overwhelming to anyone trying to pick out patterns and trends, but IBM Watson has been working with businesses and other hospitals to make this process smoother. Titte Srinivas, M.D., a transplant nephrologist and Professor of Medicine at MUSC Health, wants to incorporate this technology into the management of his kidney transplant patients by practicing data-driven care. Information about treatment history, risk factors, and outcome trends will help inform and guide their care.

"I want the electronic medical record to become the intelligent brain that guides us toward better outcomes, not what most clinicians see it as now—an expensive, passive data repository,” says Srinivas.

Watson, a commercial cognitive computing system created by IBM, was all over the news in 2011 for winning Jeopardy. Since then, the program has been modified for Health care applications. MD Anderson is using IBM Watson to help select patients for targeted gene therapies. CVS Health announced its own partnership with IBM Watson in Fortune magazine on July 29 to help Health care professionals crunch data more effectively. MUSC Health is one of only a handful of hospitals using Watson to analyze data, and it is the first to use it for transplant patients.

Watson will strengthen the Kidney Transplant Program at MUSC Health by cataloging all of its information. As a computing system, Watson can read and understand both structured and unstructured data, meaning it can process both numbers and longer, freestyle notes. Since Watson comprehends natural language, physicians no longer have to read through pages and pages of information to access the most important aspects of a patient’s health. The program can do it for them. Srinivas says it is important to look at the information we already have and analyze it. He wants to “use our past data to guide our future practice.”

The Scientific Registry of Transplant Recipients (SRTR) currently predicts various outcomes associated with kidney transplant, but it mainly captures baseline data and does not incorporate dynamically evolving information that accumulates during patient care. The predictive models that will be built using Watson will consider patient variables such as cholesterol levels, blood pressure, drug levels and dosage, and the clinical events that happen after the transplant. By improving upon the SRTR variable, physicians at MUSC Health can offer the best possible treatment based on the trends in this information. For instance, if trends in the data show that patients with particular vital signs respond best to a specific type and dosage of medication, then the transplant physician would prescribe them that medicine at an appropriate dose.

Srinivas has been working with David J. Taber, Pharm.D., of the Division of Transplant Surgery and Patrick D. Mauldin, Ph.D., of the Division of General Internal Medicine & Geriatrics to bring IBM Watson to MUSC Health, and they are confident this program can help kidney transplants be more successful. “Our ideal is that you die of natural causes 25-30 years after the transplant,” Srinivas says.

Srinivas, Taber, and Mauldin are partnering with IBM Watson to develop an alert system for patient medical records using a red, yellow, and green color system. Each aspect of the patient’s condition would be flagged according to color, enabling the physician to make an informed decision about care. A patient with high cholesterol would be flagged as red under cholesterol but could be flagged green under blood cell count, for example. The flags will be built off the model generated by Watson, and this model will be run daily as data accrues. This kind of system could help physicians quickly see what may have changed since the patient’s last visit or even consolidate the medical records of a patient who has been treated by multiple physicians. Most importantly, these alerts would trigger specific action plans to prevent the patient’s transplant being compromised. Watson will also be used to define what determines value in transplant care by modeling costs associated with outcomes.

This vision of truly data-driven care is being brought to fruition with the support of MUSC President David J. Cole, M.D., MUSC Medical Center Executive Director and Chief Executive Officer Patrick J. Cawley, M.D., MBA, former Chief Information Officer Frank Clark, Ph.D., Chief Research Information Officer Leslie (Les) A. Lenert, M.D., MS, and Chief Analytics Officer John Long.

The next step for Srinivas is to implement intervention trials to test this model of practice and see how it changes behavior. He plans to have validation results by October of this year.