A multivariate Bayesian scan statistic for early event detection and characterization

Daniel Neill, Gregory F. Cooper

Research output: Contribution to journalArticle

Abstract

We present the multivariate Bayesian scan statistic (MBSS), a general framework for event detection and characterization in multivariate spatial time series data. MBSS integrates prior information and observations from multiple data streams in a principled Bayesian framework, computing the posterior probability of each type of event in each space-time region. MBSS learns a multivariate Gamma-Poisson model from historical data, and models the effects of each event type on each stream using expert knowledge or labeled training examples. We evaluate MBSS on various disease surveillance tasks, detecting and characterizing outbreaks injected into three streams of Pennsylvania medication sales data. We demonstrate that MBSS can be used both as a "general" event detector, with high detection power across a variety of event types, and a "specific" detector that incorporates prior knowledge of an event's effects to achieve much higher detection power. MBSS has many other advantages over previous event detection approaches, including faster computation and easy interpretation and visualization of results, and allows faster and more accurate event detection by integrating information from the multiple streams. Most importantly, MBSS can model and differentiate between multiple event types, thus distinguishing between events requiring urgent responses and other, less relevant patterns in the data.

Original languageEnglish (US)
Pages (from-to)261-282
Number of pages22
JournalMachine Learning
Volume79
Issue number3
DOIs
StatePublished - Jun 1 2010

Fingerprint

Statistics
Detectors
Time series
Sales
Visualization

Keywords

  • Biosurveillance
  • Event characterization
  • Event detection
  • Scan statistics

ASJC Scopus subject areas

  • Software
  • Artificial Intelligence

Cite this

A multivariate Bayesian scan statistic for early event detection and characterization. / Neill, Daniel; Cooper, Gregory F.

In: Machine Learning, Vol. 79, No. 3, 01.06.2010, p. 261-282.

Research output: Contribution to journalArticle

@article{561fe8f33faa42ba931db8a1cbc40892,
title = "A multivariate Bayesian scan statistic for early event detection and characterization",
abstract = "We present the multivariate Bayesian scan statistic (MBSS), a general framework for event detection and characterization in multivariate spatial time series data. MBSS integrates prior information and observations from multiple data streams in a principled Bayesian framework, computing the posterior probability of each type of event in each space-time region. MBSS learns a multivariate Gamma-Poisson model from historical data, and models the effects of each event type on each stream using expert knowledge or labeled training examples. We evaluate MBSS on various disease surveillance tasks, detecting and characterizing outbreaks injected into three streams of Pennsylvania medication sales data. We demonstrate that MBSS can be used both as a {"}general{"} event detector, with high detection power across a variety of event types, and a {"}specific{"} detector that incorporates prior knowledge of an event's effects to achieve much higher detection power. MBSS has many other advantages over previous event detection approaches, including faster computation and easy interpretation and visualization of results, and allows faster and more accurate event detection by integrating information from the multiple streams. Most importantly, MBSS can model and differentiate between multiple event types, thus distinguishing between events requiring urgent responses and other, less relevant patterns in the data.",
keywords = "Biosurveillance, Event characterization, Event detection, Scan statistics",
author = "Daniel Neill and Cooper, {Gregory F.}",
year = "2010",
month = "6",
day = "1",
doi = "10.1007/s10994-009-5144-4",
language = "English (US)",
volume = "79",
pages = "261--282",
journal = "Machine Learning",
issn = "0885-6125",
publisher = "Springer Netherlands",
number = "3",

}

TY - JOUR

T1 - A multivariate Bayesian scan statistic for early event detection and characterization

AU - Neill, Daniel

AU - Cooper, Gregory F.

PY - 2010/6/1

Y1 - 2010/6/1

N2 - We present the multivariate Bayesian scan statistic (MBSS), a general framework for event detection and characterization in multivariate spatial time series data. MBSS integrates prior information and observations from multiple data streams in a principled Bayesian framework, computing the posterior probability of each type of event in each space-time region. MBSS learns a multivariate Gamma-Poisson model from historical data, and models the effects of each event type on each stream using expert knowledge or labeled training examples. We evaluate MBSS on various disease surveillance tasks, detecting and characterizing outbreaks injected into three streams of Pennsylvania medication sales data. We demonstrate that MBSS can be used both as a "general" event detector, with high detection power across a variety of event types, and a "specific" detector that incorporates prior knowledge of an event's effects to achieve much higher detection power. MBSS has many other advantages over previous event detection approaches, including faster computation and easy interpretation and visualization of results, and allows faster and more accurate event detection by integrating information from the multiple streams. Most importantly, MBSS can model and differentiate between multiple event types, thus distinguishing between events requiring urgent responses and other, less relevant patterns in the data.

AB - We present the multivariate Bayesian scan statistic (MBSS), a general framework for event detection and characterization in multivariate spatial time series data. MBSS integrates prior information and observations from multiple data streams in a principled Bayesian framework, computing the posterior probability of each type of event in each space-time region. MBSS learns a multivariate Gamma-Poisson model from historical data, and models the effects of each event type on each stream using expert knowledge or labeled training examples. We evaluate MBSS on various disease surveillance tasks, detecting and characterizing outbreaks injected into three streams of Pennsylvania medication sales data. We demonstrate that MBSS can be used both as a "general" event detector, with high detection power across a variety of event types, and a "specific" detector that incorporates prior knowledge of an event's effects to achieve much higher detection power. MBSS has many other advantages over previous event detection approaches, including faster computation and easy interpretation and visualization of results, and allows faster and more accurate event detection by integrating information from the multiple streams. Most importantly, MBSS can model and differentiate between multiple event types, thus distinguishing between events requiring urgent responses and other, less relevant patterns in the data.

KW - Biosurveillance

KW - Event characterization

KW - Event detection

KW - Scan statistics

UR - http://www.scopus.com/inward/record.url?scp=77953023188&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=77953023188&partnerID=8YFLogxK

U2 - 10.1007/s10994-009-5144-4

DO - 10.1007/s10994-009-5144-4

M3 - Article

VL - 79

SP - 261

EP - 282

JO - Machine Learning

JF - Machine Learning

SN - 0885-6125

IS - 3

ER -