Oct 12, 2012
Xerox Grenoble Works to Make Hospitals Safer, More Efficient with Imaging and Text Mining Advances
By Catherine Arnst
While the U.S. presidential candidates argue about how best to overhaul the problem-plagued U.S. healthcare system, Xerox is already doing it – developing advanced technological solutions to costly health system problems such as hospital-acquired infections and safe monitoring of premature infants. At the Xerox Research Centre Europe (XRCE) in Grenoble, France, researchers are combining advanced computer vision technologies, sophisticated algorithms and artificial intelligence techniques in ways that not only allow computers to “see,” but understand and analyze what they are seeing in ways that can make healthcare safer and more efficient.
During a day of demonstrations on Oct. 3 at XRCE for the international press and industry analysts, principal scientist Dr. Raja Bala explained how Xerox is protecting fragile, infection-prone premature infants by developing a system that constantly monitors their vital signs – without making contact. Currently, the heart rate and other vital signs of newborns in an incubator are monitored by placing electrodes on their paper-thin skin or attaching sensors to an ear or fingertip. Bala’s team has developed a system that trains an overhead video and thermal camera on the infant to obtain accurate pulse measurements, breathing and body temperature. Xerox applies patented methods that combine and analyze the video streams coming out of the two cameras to determine the baby’s vital signs. “The secret sauce is the image processing software – that’s where the real innovation comes in,” said Bala. The system is currently being tested in a hospital in India, and Bala envisions it being deployed in a number of difficult medical environments.
XRCE is also deploying text and image mining techniques against hospital acquired infections (HAI), which affect five to 10 percent of patients in Europe. HAIs cause 50,000 deaths in Europe and 99,000 in the U.S. every year. Scientists at Xerox have been working with French medical researchers for four years to develop a system that can read and analyze medical records, and search for specific terms, sequences of facts, drug names and how they are linked to determine if a hospital patient may have contracted an infection. Because each patient’s information is unique, the system must be able to sift through all of this data and place it in context to make the right call.
If the system recognizes that a patient may be at risk, it alerts hospital staff so they can treat the infection quickly, and take measures to prevent it from spreading. In one test, it analyzed 815 patient records and accurately identified HAIs 87 percent of the time. The system is currently being piloted in a French hospital with an evaluation expected for the end of the year.
Dr. Monica Beltrametti, Xerox Chief Services Research Officer and VP and Centre Manager of XRCE, said such systems are making healthcare services the fastest growing segment of Xerox’s business. “Text mining combined with visualization and artificial intelligence can revolutionize this field,” said Beltrametti.
Catherine Arnst is an award-winning freelance writer and editor specializing in medicine and science, and was previously Senior Medical writer for BusinessWeek magazine.
Note: Catherine Arnst filed this content as a paid contributor to Xerox. The content is the author’s opinion and does not necessarily reflect the views of Xerox.