Recursive least squares background prediction of univariate syndromic surveillance data
2009

Improving Outbreak Detection with Recursive Least Squares

Sample size: 16 publication Evidence: moderate

Author Information

Author(s): Najmi Amir-Homayoon, Howard Burkom

Primary Institution: The Johns Hopkins University Applied Physics Laboratory

Hypothesis

Can a recursive least squares method improve the detection of outbreaks in syndromic surveillance data?

Conclusion

The modified RLS method provides better sensitivity for detecting anomalies in biosurveillance data compared to the baseline W2 method.

Supporting Evidence

  • The modified RLS method showed improved sensitivity at multiple background alert rates.
  • Detection performance was compared to the CDC's W2 method using 16 sets of syndromic data.
  • The study found that the RLS method consistently outperformed the W2 method, especially for weak signals.

Takeaway

This study shows a new way to predict health data trends that helps find outbreaks faster, using a special math method.

Methodology

The study used an adaptive recursive least squares method to analyze syndromic data and correct for weekly patterns.

Limitations

The method requires a sufficient scale of time series data and a significant amount of historical data for effective predictions.

Participant Demographics

Data from military outpatient clinics and civilian physician office visits in large metropolitan areas.

Digital Object Identifier (DOI)

10.1186/1472-6947-9-4

Want to read the original?

Access the complete publication on the publisher's website

View Original Publication