The hospital's informatics group, a research organization separate from the IT department, designed the system with grant money from the National Institutes of Health. The goal was to create a forecasting tool to improve the preparedness of the emergency room and clinical departments, says CTO Scott Ogawa. The hospital could use the forecasts when setting staffing levels and purchasing drugs and supplies.
"If you think you may have an outbreak of GI gastrointestinal symptoms during a certain month, you might staff more housekeeping," Ogawa says. The system might also predict more orthopedic cases during the spring, when more people are playing outdoor sports in the Northeast, or more respiratory cases during flu season.
The system takes advantage of off-the-shelf data warehousing software that hospital officials would not name and employs ARIMA (AutoRegressive Integrated Moving Average) modeling, a statistical technique that constantly updates its forecasts based on new information, according to CIO Dan Nigrin, who in addition to being a medical doctor holds a master's degree in medical informatics. ARIMA modeling is popular in the financial-services sector, Nigrin says. (For more on ARIMA, see mathstat.carleton.ca/ ~help/sashtml/ets/chap7/sect1.htm.)
Children's was just about to pilot the system and publish the results in a medical journal when terrorists attacked the World Trade Center and the Pentagon. "When 9-11 came, the relevance of this stuff for bioterrorism was obvious," Nigrin says.
Two developers, one of them a medical doctor, spent four months creating an intuitive front end and making the system production-ready.