1. Methods for Real-Time Detection and Assessment of Disease Outbreaks Using Information Technology Michael Wagner, M.D., Ph.D. Director, Real-Time Outbreak and Disease Surveillance Laboratory Assistant Professor, Medicine and Intelligent Systems Center for Biomedical Informatics University of Pittsburgh Pittsburgh, PA

2. What is the Mission of the RODS Lab? First Hint of Trouble statistical analysis of data astute observer definitive diagnosis of new or “terrorism” organism Analysis/Characterization Is it an emergency? Quarantine? Get more antibiotics? Additional data collection ”shoeleather” microchip testing decision support at the point-of-care RESPONSES

3.  Every product already has a UPC bar code  Every purchase already is scanned optically  12 big chains already merged thousands of stores and already receive daily batch feeds of sales data from those stores by midnight  We “asked” for the data  We worked with the industry to securely transmit the data every day to Pittsburgh (by 3 pm)  We built the databases, created the analytic product categories, and the analytic tools National Retail Data Monitor

4. Current Status 18,000 stores (35% market share) 220 user accounts in 33 States, CDC

5. Future Plans  Achieve 70% market share  Decrease time latency  Add monitoring of prescription antibiotics  Deploy in second country  More automation of detection analysis  Transition NRDM development supported by PA Bioinformatics Grant #ME-01-737, The Alfred P. Sloan Foundation and New York State

6. For More Information  www.health.pitt.edu/rods www.health.pitt.edu/rods  Free account: nrdm-accounts@cbmi.pitt.edu nrdm-accounts@cbmi.pitt.edu (public health officials only)  Paper: Wagner, Robinson, Tsui, Espino, Design of a National Retail Data Monitor, JAMIA, 2003;10(5) 409-20 NRDM development supported by PA Bioinformatics Grant #ME-01-737, The Alfred P. Sloan Foundation and New York State

7. Real-time Surveillance of Hospital Data: The RODS System

8. HL7 Admission, Discharge, Transfer Message MSH|^~\&||xxx||RODS|200202241715||ADT^A04|2002022XX XXXXXX|P|2.3 PID|||||||^020|M|||^^^^84204||||| PV1||E|||||||||||||||||98765432|||||||||||||||||||| ||||200202151830|| DG1||||SORE THROAT,COUGH IN1||||||||||||||||||||||||||||||||||||||||||||^^^^ 84056 Zip code Visit date and time Free-text chief complaint

9. “N/V/D” Chief Complaint Naïve Bayes Classifier Naïve Bayes Text Classifier P(Respiratory|NVD)=.05 P(Botulinic|NVD)=.001 P(Constitutional|NVD)=.01 P(GI|NVD) =.9 P(Hemorrhagic|NVD)=.001 P(Neurologic|NVD)=.001 P(Rash|NVD)=.001 P(None|NVD)=.036 Output GI Prodrome

10. Copyright University of Pittsburgh 2002 All visitsRespiratoryGIRash Botulinic NeurologicalHemorrhagicConstitutional Time Series

11. Map

12. RODS Open Source Project  Real-time biosurveillance depends (heavily) on software  Good software doesn’t grow on trees  University of Pittsburgh created RODS and then released it—for free  Consultants, users, programmers also do not arise by spontaneous generation  Commonwealth of Pennsylvania funding is being used to catalyze and transition RODS to become one of the world’s most used and most advanced syndromic surveillance systems

13. For More Information  www.health.pitt.edu/rods  RODS Open Source Project: http://openrods.sourceforge.net  Papers: Tsui et al. Technical Description of RODS. JAMIA, 2003;10(5) 409-20 RODS development supported by PA Bioinformatics Grant #ME-01-737, AHRQ, National Library of Medicine Training Grant (Dr. Espino)