IBM Research streaming analytics solution saves time and lives     

IBM Research streaming analytics solution saves time and lives - overview

Medical researchers at Columbia University and Columbia Presbyterian Hospital are discovering that streaming analytics technology developed at IBM Research can help detect complications in brain-injured patients. Initial findings have led the researchers to believe that they might spot signs of brain injury up to two days earlier than they would by using traditional methods. Ultimately, doctors would have treatment options where none currently exist.

One of the most typical and severe complications for patients who suffer a bleeding stroke from a ruptured brain aneurysm is delayed ischemia. This is a life-threatening condition in which the brain does not get enough blood. As of now, detectable symptoms appear only after blood flow has been reduced significantly, forcing medical professionals to apply reactive measures instead of preventive treatment. One-fifth of patients suffering from impending stroke show no symptoms at all, and doctors frequently intervene too late to stave off the worst consequences.

Made in IBM laboratories

IBM streaming analytics analyzes large volumes of "data in motion." Using this technology, researchers at Columbia University believe they can uncover the patterns in symptom progression not visible to the naked eye, and spot the onset of the condition up to 48 hours earlier than current methods.

The detection of these early warning signs would let doctors begin treatments sooner or thwart the progression of the condition altogether.

"The ability to analyze massive volumes of medical data to uncover early warning signs for this potentially devastating and life-threatening complication could lead to significant improvements in how this condition is treated," says Dr. Michael Schmidt, director of neuromonitoring and informatics, Neurological Intensive Care Unit Columbia University.

"We need the ability to not only uncover the hidden data patterns in the lab, but then take what we learn and use it in real time at the bedside for the benefit of the patient," says Dr. Schmidt. "This is the goal of translational research and applying IBM streaming analytics is an innovative and elegant approach to achieving real time complex analysis of bedside patient information in the intensive care setting."

Multi-phase analysis and instantaneous insights

The first phase of the project involves uncovering patterns within the volumes of data associated with the patient's complications. Using analytics software, the researchers process physiological data streams such as EEG feeds, blood pressure, blood oxygen levels and temperature readings in conjunction with data such as lab test results, patient information and symptoms reported by medical professionals and patients. Analyzing this information may help physicians uncover hidden patterns in various test results that would otherwise be impossible to correlate.

Once the key hallmarks of this life-threatening condition are identified and validated, the second phase of the project will let researchers move the project to the neurological intensive care unit (Neuro ICU). There they can gather data in real time from patients, testing the warning signs previously identified and offering medical professionals instantaneous insights into the condition of their patients.

Medical professionals who work in a Neuro ICU measure and assess more than 200 variables when evaluating a patient. From heart rates, temperature, blood pressure, brain and heart activity readings, they must evaluate a constant and complex stream of data. IBM's breakthrough streaming analytics software, IBM Streaming Analytics, combines data from medical tests and equipment in the Neuro ICU with data from other sources such as databases and images. The software analyzes all of this data in real time, giving medical professionals instant updates on a patient's condition, spotting trends and symptoms as they emerge.

Only the beginning

"We are only at the beginning of what is possible with streaming analytics to detect complications in stroke patients and to intervene early during heart attacks to minimize cardiac muscle damage," says Nagui Halim, director and research scientist, IBM Streaming Analytics. "It is not only telecommunications companies in Asia and government agencies in Europe that are benefiting from IBM streaming analytics. This technology is helping healthcare organizations use information in ways they never thought possible."

A version of this article originally appeared on the IBM intranet. It was written by Elliot Luber.

Artwork: Jane Luongo

Last updated on November 1, 2010

Text is best viewed in Firefox or Safari.