For the NeoNatal Intensive Care unit at the Hospital for Sick Children (SickKids) in Toronto, big data tools have proven valuable to allow doctors to monitor the vital signs of premature infants around the clock.
The hospital is working on early identification of newborn babies with an infection called late-onset neonatal sepsis, a blood infection that occurs in children between days 8 and 89.
“We’ve been focusing on the early identification of what we call late-onset neonatal sepsis [LONS] infections in newborn babies more than 72 hours after they’ve been admitted into a neonatal unit,” Dr. Andrew James, head of the Neonatal Intensive Care Unit at Toronto’s Hospital for Sick Children, told eWEEK. “The babies are very immature, and natural defense mechanisms are underdeveloped,” he noted.
Doctors get a digital reading on respiratory rates, heart rates, blood pressure and blood oxygen saturation, and can analyze this data using a platform called Artemis. It consists of three servers as well as IBM’s InfoSphere Streams software and allows doctors to monitor infants’ vital signs in real time and detect changes in their conditions.
The hospital also uses Philips IntelliVue MP70 monitors to track babies’ vital signs.
InfoSphere Streams is “software that’s been developed to run complex algorithms on modal streams of data,” said James, who collaborated on Artemis with Dr. Carolyn McGregor, Canada Research Chair in Health Informatics at the University of Ontario Institute of Technology.
“Our strategy is to monitor these data streams, and we have an algorithm that examines the data streams looking for features that are known to occur before the infection becomes clinically apparent,” James said.
Previously, neonatologists had difficulty keeping track of infants’ vital data manually as well as from medical monitoring equipment.
The Toronto hospital uses algorithms based on Streamed Programming Language (SPL), a computing language that IBM developed to provide the instructions that InfoSphere Streams requires to run. Instructions in the algorithm run the analysis and can form a new stream of data, James said.
The project began with three laptops and then moved to three servers to handle more data, James said.
“We’re gradually moving up to 40 bed spaces so we needed more computing power,” he said. “When we were gathering data from eight babies, we were gathering 90 million data points per day.”
To compile data from 40 bed spaces, the hospital would need to store 5 terabytes of data regularly, he said.
By using the Artemis framework, SickKids aims to reduce the time it takes a clinician to assess health data from multiple sources and allow clinicians to control the rules engine of Artemis to improve care environments. In addition, Artemis will provide clinical alerts to both synchronous and asynchronous data, and process infants’ data in real time, according to the hospital.
The hospital hopes that data from Artemis will allow neonatologists to predict the presence of infections that cause death in infants and bring about a quicker intervention and better outcomes, according to a research report by the University of Toronto, University of Ontario Institute of Technology and the IBM T.J. Watson Research Center.
With the data, doctors could start infants on antibiotics before they become ill, James told CNBC for a Sept 18 documentary “Rise of the Machines.” Almost every second, each infant generates more than 1,200 points of data and nearly 90 million points of data per day, and the data is often ignored, according to the broadcast. James noted that the human brain is incapable of processing this information without the help of big data algorithms. “The human brain just doesn’t have the capacity to analyze it,” he told CNBC.
More information will help doctors “hone in on the problem,” McGregor said in the documentary.