Limits...
Disease surveillance using a hidden Markov model.

Watkins RE, Eagleson S, Veenendaal B, Wright G, Plant AJ - BMC Med Inform Decis Mak (2009)

Bottom Line: At false alarm rates around 0.01, the HMM algorithms provided the best overall outbreak detection performance, having higher sensitivity than the cusum-based Methods and a generally shorter time to detection for larger outbreaks.Overall, the 14-day HMM had a significantly greater area under the receiver operator characteristic curve than the EARS C3 and 7-day negative binomial cusum algorithms.Our findings suggest that the HMM provides an effective method for the surveillance of sparse small area notifiable disease data at low false alarm rates.

View Article: PubMed Central - HTML - PubMed

Affiliation: Curtin Health Innovation Research Institute, Curtin University of Technology, Perth, Australia. Rochelle.Watkins@curtin.edu.au

ABSTRACT

Background: Routine surveillance of disease notification data can enable the early detection of localised disease outbreaks. Although hidden Markov models (HMMs) have been recognised as an appropriate method to model disease surveillance data, they have been rarely applied in public health practice. We aimed to develop and evaluate a simple flexible HMM for disease surveillance which is suitable for use with sparse small area count data and requires little baseline data.

Methods: A Bayesian HMM was designed to monitor routinely collected notifiable disease data that are aggregated by residential postcode. Semi-synthetic data were used to evaluate the algorithm and compare outbreak detection performance with the established Early Aberration Reporting System (EARS) algorithms and a negative binomial cusum.

Results: Algorithm performance varied according to the desired false alarm rate for surveillance. At false alarm rates around 0.05, the cusum-based algorithms provided the best overall outbreak detection performance, having similar sensitivity to the HMMs and a shorter average time to detection. At false alarm rates around 0.01, the HMM algorithms provided the best overall outbreak detection performance, having higher sensitivity than the cusum-based Methods and a generally shorter time to detection for larger outbreaks. Overall, the 14-day HMM had a significantly greater area under the receiver operator characteristic curve than the EARS C3 and 7-day negative binomial cusum algorithms.

Conclusion: Our findings suggest that the HMM provides an effective method for the surveillance of sparse small area notifiable disease data at low false alarm rates. Further investigations are required to evaluation algorithm performance across other diseases and surveillance contexts.

Show MeSH

Related in: MedlinePlus

Sensitivity of the outbreak detection algorithms according to false alarm rates less than 0.1 for the smaller less clustered simulation scenario (S1).
© Copyright Policy - open-access
Related In: Results  -  Collection

License
getmorefigures.php?uid=PMC2735038&req=5

Figure 3: Sensitivity of the outbreak detection algorithms according to false alarm rates less than 0.1 for the smaller less clustered simulation scenario (S1).

Mentions: The sensitivity of the algorithms for false alarm rates less than 0.1 is summarised for the smaller less clustered and larger more clustered simulation scenarios in Figures 3 and 4 respectively. The sensitivity of the 28-day HMM is not displayed in Figures 3 and 4 as it was indistinguishable from that of the 14-day HMM with a prior mean of 3. The shorter-baseline HMMs were more sensitive when the smaller outbreak prior mean was used; however, there was no appreciable difference for both prior means tested for the 28-day HMM. The following results present the performance of the 7, 14 and 28-day HMMs using the outbreak state prior means of 2, 2, and 3 respectively.


Disease surveillance using a hidden Markov model.

Watkins RE, Eagleson S, Veenendaal B, Wright G, Plant AJ - BMC Med Inform Decis Mak (2009)

Sensitivity of the outbreak detection algorithms according to false alarm rates less than 0.1 for the smaller less clustered simulation scenario (S1).
© Copyright Policy - open-access
Related In: Results  -  Collection

License
Show All Figures
getmorefigures.php?uid=PMC2735038&req=5

Figure 3: Sensitivity of the outbreak detection algorithms according to false alarm rates less than 0.1 for the smaller less clustered simulation scenario (S1).
Mentions: The sensitivity of the algorithms for false alarm rates less than 0.1 is summarised for the smaller less clustered and larger more clustered simulation scenarios in Figures 3 and 4 respectively. The sensitivity of the 28-day HMM is not displayed in Figures 3 and 4 as it was indistinguishable from that of the 14-day HMM with a prior mean of 3. The shorter-baseline HMMs were more sensitive when the smaller outbreak prior mean was used; however, there was no appreciable difference for both prior means tested for the 28-day HMM. The following results present the performance of the 7, 14 and 28-day HMMs using the outbreak state prior means of 2, 2, and 3 respectively.

Bottom Line: At false alarm rates around 0.01, the HMM algorithms provided the best overall outbreak detection performance, having higher sensitivity than the cusum-based Methods and a generally shorter time to detection for larger outbreaks.Overall, the 14-day HMM had a significantly greater area under the receiver operator characteristic curve than the EARS C3 and 7-day negative binomial cusum algorithms.Our findings suggest that the HMM provides an effective method for the surveillance of sparse small area notifiable disease data at low false alarm rates.

View Article: PubMed Central - HTML - PubMed

Affiliation: Curtin Health Innovation Research Institute, Curtin University of Technology, Perth, Australia. Rochelle.Watkins@curtin.edu.au

ABSTRACT

Background: Routine surveillance of disease notification data can enable the early detection of localised disease outbreaks. Although hidden Markov models (HMMs) have been recognised as an appropriate method to model disease surveillance data, they have been rarely applied in public health practice. We aimed to develop and evaluate a simple flexible HMM for disease surveillance which is suitable for use with sparse small area count data and requires little baseline data.

Methods: A Bayesian HMM was designed to monitor routinely collected notifiable disease data that are aggregated by residential postcode. Semi-synthetic data were used to evaluate the algorithm and compare outbreak detection performance with the established Early Aberration Reporting System (EARS) algorithms and a negative binomial cusum.

Results: Algorithm performance varied according to the desired false alarm rate for surveillance. At false alarm rates around 0.05, the cusum-based algorithms provided the best overall outbreak detection performance, having similar sensitivity to the HMMs and a shorter average time to detection. At false alarm rates around 0.01, the HMM algorithms provided the best overall outbreak detection performance, having higher sensitivity than the cusum-based Methods and a generally shorter time to detection for larger outbreaks. Overall, the 14-day HMM had a significantly greater area under the receiver operator characteristic curve than the EARS C3 and 7-day negative binomial cusum algorithms.

Conclusion: Our findings suggest that the HMM provides an effective method for the surveillance of sparse small area notifiable disease data at low false alarm rates. Further investigations are required to evaluation algorithm performance across other diseases and surveillance contexts.

Show MeSH
Related in: MedlinePlus