Jove
Visualize
Contact Us
JoVE
x logofacebook logolinkedin logoyoutube logo
ABOUT JoVE
OverviewLeadershipBlogJoVE Help Center
AUTHORS
Publishing ProcessEditorial BoardScope & PoliciesPeer ReviewFAQSubmit
LIBRARIANS
TestimonialsSubscriptionsAccessResourcesLibrary Advisory BoardFAQ
RESEARCH
JoVE JournalMethods CollectionsJoVE Encyclopedia of ExperimentsArchive
EDUCATION
JoVE CoreJoVE BusinessJoVE Science EducationJoVE Lab ManualFaculty Resource CenterFaculty Site
Terms & Conditions of Use
Privacy Policy
Policies

Related Experiment Videos

Automated time series forecasting for biosurveillance.

Howard S Burkom1, Sean Patrick Murphy, Galit Shmueli

  • 1The Johns Hopkins University Applied Physics Laboratory, MD, USA. howard.burkom@jhuapl.edu

Statistics in Medicine
|March 6, 2007
PubMed
Summary
This summary is machine-generated.

Related Concept Videos

You might also read

Related Articles

Articles linked to this work by shared authors, journal, and citation graph.

Sort by
Same author

Deployment to Karshi-Khanabad Air Base, Uzbekistan between 2001 and 2005 and subsequent risk of specific cancers among US service members.

Journal of the National Cancer Institute·2026
Same author

To Explain, to Predict, or to Describe: Figuring out the Study Goal [Commentary on "On the Uses and Abuses of Regression Models" by Carlin and Moreno-Betancur].

Statistics in medicine·2025
Same author

Atomist or holist? A diagnosis and vision for more productive interdisciplinary AI ethics dialogue.

Patterns (New York, N.Y.)·2023
Same author

Improving the Prediction of Persistent High Health Care Utilizers: Retrospective Analysis Using Ensemble Methodology.

JMIR medical informatics·2022
Same author

Assessing the Value of Unsupervised Clustering in Predicting Persistent High Health Care Utilizers: Retrospective Analysis of Insurance Claims Data.

JMIR medical informatics·2021
Same author

Discussion on "Assessing the goodness of fit of logistic regression models in large samples: A modification of the Hosmer-Lemeshow test" by Giovanni Nattino, Michael L. Pennell, and Stanley Lemeshow.

Biometrics·2020

The Holt-Winters method best predicted syndromic data trends for biosurveillance, outperforming regression models. This improves anomaly detection by removing systematic data behavior for more robust monitoring.

Area of Science:

  • Epidemiology
  • Biostatistics
  • Time Series Analysis

Background:

  • Biosurveillance relies on control charts for anomaly detection.
  • Traditional methods require data free of trends and systematic patterns.
  • Time series forecasting can preprocess data by removing systematic behavior.

Purpose of the Study:

  • To compare the predictive accuracy of three time series forecasting methods for biosurveillance.
  • To evaluate the effectiveness of these methods in removing systematic behavior from syndromic data.

Main Methods:

  • Three forecasting methods were compared: non-adaptive regression, adaptive regression, and Holt-Winters exponential smoothing.
  • Predictive accuracy was assessed using root-mean-square error, median absolute per cent error (MedAPE), and median absolute deviation.

Related Experiment Videos

  • The methods were applied to 16 authentic syndromic data streams without individual series tuning.

    • Main Results:

    • The Holt-Winters method demonstrated superior performance based on median-based error metrics (MedAPE averaged 9.7).
    • Adaptive regression (MedAPE 11.6) outperformed non-adaptive regression (MedAPE 16.5), which was sensitive to baseline data changes.
    • Holt-Winters was most effective in reducing serial autocorrelation, with most 1-day-lag coefficients below 0.15.

    Conclusions:

    • The Holt-Winters method is recommended for preprocessing biosurveillance data due to its accuracy and ability to handle data complexities.
    • Tuning forecasting models to individual series can further enhance prediction, but requires reliable data classification for practical implementation.