Related Experiment Videos
Factors affecting automated syndromic surveillance.
Ling Wang1, Marco F Ramoni, Kenneth D Mandl
1Department of Biostatistics, Boston University School of Public Health, 715 Albany Street, Boston, MA 02118, USA.
Artificial Intelligence in Medicine
|July 19, 2005
Summary
An automated system using syndromic data accurately detects outbreaks with 84.8% true detection accuracy. Integrating multiple data sources enhances public health surveillance and early anomaly identification.
Area of Science:
- Public Health
- Epidemiology
- Biostatistics
Background:
- Bioterrorism and epidemics necessitate advanced public health surveillance.
- Early detection of anomalies in public health data is crucial.
Purpose of the Study:
- To develop and evaluate an automated outbreak detection system using syndromic data.
- To assess the system's performance in identifying respiratory syndrome outbreaks.
Main Methods:
- Utilized an autoregressive model with seasonal components for online monitoring of daily chief complaints.
- Evaluated system performance using real data to estimate false positive rates.
- Assessed true positive rates by injecting simulated outbreaks of varying shapes and sizes.
- Employed directed graphical models to analyze the impact of exogenous factors.
Main Results:
- Achieved 84.8% true detection accuracy for week-long outbreaks across different shapes.
- Outbreak size impacts the earliness of detection.
- Exogenous factors influence false and true positive rates, offering potential for accuracy improvement.
Conclusions:
- The developed automated system demonstrates significant potential for syndromic surveillance.
- Integrating multiple data sources can substantially enhance the accuracy of outbreak detection systems.